prof_pic.jpg

I am an Assistant Professor at KAIST Graduate School of AI, where I direct the Structured and Probabilistic Machine Learning (SPMLSPML) Lab.

My research focuses on structured, probabilistic machine learning with applications to drug discovery and material design. I enjoy bringing novel and practical perspective to scientific problems and diving deep into the underlying physics, chemistry, and biology.

CV

Members

Yunhui Jang, Hyosoon Jang, Hyomin Kim, Seonghyun Park, Seongsu Kim, Kiyoung Seong, Minkyu Kim, Dongyeop Woo, Nayoung Kim, Taewon Kim, Hyunjin Seo, Yinhua Piao (Postdoc), Yoonho Kim, Honghui Kim (Postdoc), Seonghwan Kim (Postdoc)

Alumni: Haeji Ko, Juwon Hwang, Minsu Kim

Research

We develop structured and probabilistic machine learning to infer, predict, and design molecular and material systems. Our goal is to expand what scientists can learn from experiments and simulations, and what they can investigate with that knowledge.

Partition-preserving factor transformations Bias-sensitive auxiliary models Local search improvement operators Search results as training targets Sparse graph sequence encodings Local signals for credit assignment Sampler training without target samples Replay of exploratory samples Offline design with learned property proxies Explicit molecular structure from SMILES Propose, critique, and select with agents Reduced coordinates for crystal generation Periodic structure factorization Generate internal geometry instead of rigid blocks Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Proteins & genomics Cells & perturbations Cells & perturbations Materials Materials Materials Materials Materials Materials Materials Electronic structure Electronic structure Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery Molecules & drug discovery General ML: graphical models General ML: graphical models General ML: graphical models General ML: graphical models General ML: graphical models General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML General ML scTrilemma: Balancing Identity, Invariance, and Reconstruction in Single-Cell Representation Learning · Yunhak Oh, Yoonho Lee, Junseok Lee, Namkyeong Lee, Sang-Yeon Hwang, Yinhua Piao, Hyomin Kim, Seonghwan Kim, Jaechang Lim, Woo Youn Kim, Sungsoo Ahn, Chanyoung Park · NeurIPS 2026 · Single-cell RNA-seq representation learning is fundamentally label-free: cell identities, states, and contexts are not fixed training targets, so what constitutes signal or nuisance is analysis-dependent. A single representation must therefore preserve biological identity and state, remain robust to nuisance context, and retain the gene-level variation needed for expression analysis, three demands we call the representation trilemma. To tackle this problem, we introduce scTrilemma, a latent-bottleneck VAE that routes expression-derived variation to the embedding, the decoder, or the prior rather than forcing all of it through one embedding. It gates gene tokens by expression, routes the cell representation through the decoder, and conditions the prior on unlabeled pseudo-bulk context, under a single reconstruction objective and without target annotations or auxiliary representation losses. In release-based zero-shot evaluation on successive CZ CELLxGENE Census releases, scTrilemma leads all three demands at once and preserves biological-state, differential-expression, and pathway structure across multiple disease settings. Latent interventions further show that context can be removed at almost no cost to the other demands, leaving identity against fidelity as the remaining tension. Code is publicly available at https://github.com/yunhak0/scTrilemma . scTrilemma NeurIPS 2026 Packora: Systematic Design for Generative Molecular Crystal Structure Prediction · Nayoung Kim, Kiyoung Seong, Sungsoo Ahn · arXiv 2026 · Molecular crystal structure prediction (CSP) is important in pharmaceuticals, agrochemicals, and organic electronics, where subtle differences in molecular conformation and packing can strongly affect material properties. We present Packora, a flow-based generative model for molecular CSP that jointly predicts atomic coordinates and the lattice from molecular graphs. Packora supports multi-component and organometallic crystals and can condition on any subset of molecular conformers, stereochemical labels, and space-group information within a single model. Inspired by the CCDC CSP blind test, we evaluate generation and ranking separately, using generation to isolate generator quality and ranking to measure end-to-end performance under a common relaxation and ranking pipeline. We also systematically study architecture, training, conditioning, inference, and scaling, identifying an effective design based on cacheable pairwise reasoning, training objective and numerical solver choices, conditioning dropout, and balanced scaling of pairwise and single representations. Packora outperforms the baselines on both structure generation and ranking benchmarks, achieving the best matched-budget coverage across all six generation benchmarks, as well as higher experimental-form recovery, lower experimental-form ranks, and faster convergence in ranking. Packora arXiv 2026 Discovering Crystal Structure Prediction Algorithms with an AI Co-Scientist · Kiyoung Seong, Nayoung Kim, Sungsoo Ahn · arXiv 2026 · We introduce Human-AI Co-discovery system (HACO) for scientific algorithm discovery through cross-domain search and sparse human steering. Starting from the goal of generating crystal structures from chemical compositions, HACO searched across generative modeling methodologies from multiple fields and identified MaskGIT, a masked generative model from vision, as a promising framework for crystal structure prediction (CSP). HACO instantiated this masked formulation as a discrete token model of crystal structure; guided by sparse high-level human objectives, it then added crystallographic symmetry tokens, space group stratified sampling for polymorph coverage, and sub-bin coordinate refinement, yielding the Masked Generative Crystal Transformer (MaskGXT). On the MP-20 polymorph split, MaskGXT reaches 79.06% match-everyone-to-reference (METRe) accuracy, compared with 70.87% for the strongest evaluated baseline. MaskGXT also attains the best match rate on standard MP-20 and MPTS-52 CSP benchmarks. These results provide evidence that, in domains offering cheap, fast, and well-aligned validation, transfer-guided interactive AI co-scientists can contribute to scientific algorithm discovery by identifying transferable modeling principles and combining them with targeted human domain guidance. MaskGXT / HACO arXiv 2026 MADField: Multi-fidelity Amortized Density Field for Adsorption in Nanoporous Materials · Yoonho Kim, Seongsu Kim, Sungsoo Ahn, Honghui Kim · arXiv 2026 · High-throughput computational screening of nanoporous materials for gas storage and separation requires fast and accurate characterization of adsorption equilibrium. Particle-based grand canonical Monte Carlo (GCMC) and density-based classical density functional theory (cDFT) provide simulation-based estimates of gas uptake and adsorbate density fields, but their speed-accuracy tradeoff remains insufficient for large-scale screening. In this work, we address this gap with Multi-fidelity Amortized Density Field for Adsorption in Nanoporous Materials (MADField), which reframes adsorption prediction as equilibrium density-field estimation. MADField learns from two complementary fidelities, combining broad and scalable cDFT density supervision with higher-fidelity GCMC density labels, and recovers gas uptake by integrating the predicted density field. MADField improves uptake accuracy over the strongest baselines by 6.0x for cDFT and 15.4x for GCMC, and its predicted fields accelerate cDFT solvers with 2.0x fewer iteration steps while recovering 42 percent of cases that fail under standard settings. Finally, we evaluate MADField for conventional CH4 working capacity screening on the 270k-structure ARC-MOF database. Within this space of extremely rare high-capacity targets, 167 in total, the model achieves 56x higher average precision than the strongest baseline and accelerates inference by five orders of magnitude compared to GCMC. By prioritizing the MADField rankings, selecting the top 1.7 percent of candidates recovers 95 percent of all targets, while the top 6 percent ensures 100 percent recall. MADField arXiv 2026 Atom-level Protein Representation Learning Improves Protein Structure Prediction · Taewon Kim, Hyosoon Jang, Hyunjin Seo, Seonghwan Seo, Hyeongwoo Kim, Wonho Zhung, Mingyeong Shin, Woo Youn Kim, Sungsoo Ahn · NeurIPS 2026 · Recent advances in generative modeling show that pretrained representations can improve generation as conditioning features or alignment targets. Motivated by this, we study protein representations for predicting structures beyond conventional function annotation. We propose TriProRep, a structure-aware pretraining method that jointly models three aligned residue-level views: amino-acid identity, backbone geometry, and local full-atom geometry, discretely encoded via VQ-VAE tokenizers. By pretraining to recover original tokens from generator-corrupted views, TriProRep learns to distinguish plausible but incorrect cross-view augmentations from the original protein. We further introduce RepSP, a benchmark for evaluating protein representations in structure-predictive settings. RepSP tests three uses of representations: homodimer co-folding from apo-chain representations, residue-level prediction of homodimer-derived interaction properties, and representation-aligned monomer structure prediction. Across these tasks, TriProRep improves over sequence-only and prior structure-aware representation models, while maintaining competitive performance on conventional benchmarks. TriProRep NeurIPS 2026 VibeProteinBench: An Evaluation Benchmark for Language-interfaced Vibe Protein Design · Hyunjin Seo, Hongjoon Ahn, Jimin Park, Sungjun Han, Gyubok Lee, Soojung Yang, Joseph S Brown, Leo Chen, Gina El Nesr, Feyisayo Eweje, Sarah Gurev, Hyejin Lee, Cheng-Hao Liu, Junlang Liu, Zhihui Qi, Gyu Rie Lee, Sungsoo Ahn, Jamin Shin, Sangwon Jung · arXiv 2026 · Protein design aims to compose amino-acid sequences that fold into stable three-dimensional structures while satisfying targeted functional properties. The field is increasingly shifting toward vibe protein design, where a single model is expected to generate novel sequences, engineer existing proteins, and reason about protein characteristics through flexible natural-language constraints. However, existing benchmarks typically evaluate isolated aspects of protein design or assume predefined structured inputs, making them ill-suited to assess this broad, open-ended setting. To address this gap, we present Vibe Protein Design Benchmark (VibeProteinBench), a language-interfaced benchmark that evaluates whether a single model can operate across three complementary stages of a computational protein-design workflow: recognition, engineering, and generation. Each stage is grounded in expert-curated mechanistic rationales and multi-faceted in silico validation, to computationally verify whether model outputs are biologically plausible. Across diverse general-purpose and domain-specialized LLMs, evaluated with and without tool access, no model performs strongly across all three stages at once. Beyond measuring end-task success, VibeProteinBench exposes where current systems break: we identify recurring failure patterns and characterize how tool-augmented agents select tool arguments. Together, these findings show that vibe protein design remains a substantial open challenge. VibeProteinBench arXiv 2026 Multimodal Crystal Flow: Any-to-Any Modality Generation for Unified Crystal Modeling · Kiyoung Seong, Sungsoo Ahn, Sehui Han, Changyoung Park · ICML 2026 · Crystal modeling spans a family of conditional and unconditional generation tasks, including crystal structure prediction (CSP) and de novo generation (DNG). While recent deep generative models have shown promising performance, they remain largely task-specific, lacking a unified framework that shares crystal representations across tasks. To address this limitation, we propose Multimodal Crystal Flow (MCFlow), a unified multimodal flow model that realizes multiple crystal generation tasks as distinct inference trajectories via independent time variables for atom types and crystal structures. To enable multimodal flow in a standard transformer model, we introduce a composition- and symmetry-aware atom ordering with hierarchical permutation augmentation, injecting compositional and crystallographic priors without explicit structural templates. Experiments on the MP-20 and MPTS-52 benchmarks show that a single MCFlow model is competitive with task-specific baselines across CSP, DNG, and structure-conditioned atom type generation. MCFlow ICML 2026 Learning Adaptive Perturbation-Conditioned Contexts for Robust Transcriptional Response Prediction · Yinhua Piao, Hyomin Kim, Seonghwan Kim, Yunhak Oh, Junhyeok Jeon, Sang-Yeon Hwang, Jaechang Lim, Woo Youn Kim, Chanyoung Park, Sungsoo Ahn · ICML 2026 · Predicting high-dimensional transcriptional responses to genetic perturbations is challenging because signals are sparse and experimental noise is severe. Existing methods often suffer from mean collapse, achieving high correlation by predicting the global average expression rather than perturbation-specific responses, which yields false positives and poor interpretability. Methods that add biological knowledge graphs typically treat them as dense, static priors shared across perturbations, propagating noise. We propose AdaPert, which counters mean collapse by extracting a sparse, perturbation-specific subgraph via differentiable node selection, then suppressing spurious variation in non-responsive genes while emphasizing differentially expressed ones. Across multiple benchmarks, AdaPert outperforms existing baselines, with the largest gains on DEG-aware metrics. AdaPert ICML 2026 Machine Learning Hamiltonians are Accurate Energy-Force Predictors · Seongsu Kim, Chanhui Lee, Yoonho Kim, Seongjun Yun, Honghui Kim, Nayoung Kim, Changyoung Park, Sehui Han, Sungbin Lim†, Sungsoo Ahn† · ICML 2026 · Recently, machine learning Hamiltonian (MLH) models have gained traction as fast approximations of electronic structures such as orbitals and electron densities, while also enabling direct evaluation of energies and forces from their predictions. However, despite their physical grounding, existing Hamiltonian models are evaluated mainly by reconstruction metrics, leaving it unclear how well they perform as energy-force predictors. We address this gap with a benchmark that computes energies and forces directly from predicted Hamiltonians. Within this framework, we propose QHFlow2, a state-of-the-art Hamiltonian model with an SO(2)-equivariant backbone and a two-stage edge update. QHFlow2 achieves 40% lower Hamiltonian error than the previous best model with fewer parameters. Under direct evaluation on MD17/rMD17, it is the first Hamiltonian model to reach NequIP-level force accuracy while achieving up to 20× lower energy MAE. On QH9, QHFlow2 reduces energy error by up to 20× compared to MACE. Finally, we demonstrate that QHFlow2 exhibits consistent scaling behavior with respect to model capacity and data, and that improvements in Hamiltonian accuracy effectively translate into more accurate energy and force computations. QHFlow2 ICML 2026 Riemannian MeanFlow · Dongyeop Woo, Marta Skreta, Seonghyun Park, Kirill Neklyudov†, Sungsoo Ahn† · ICML 2026 · Diffusion and flow models have become the dominant paradigm for generative modeling on Riemannian manifolds, with successful applications in protein backbone generation and DNA sequence design. However, these methods require tens to hundreds of neural network evaluations at inference time, which can become a computational bottleneck in large-scale scientific sampling workflows. We introduce Riemannian MeanFlow (RMF), a framework for learning flow maps directly on manifolds, enabling high-quality generations with as few as one forward pass. We derive three equivalent characterizations of the manifold average velocity (Eulerian, Lagrangian, and semigroup identities), and analyze parameterizations and stabilization techniques to improve training on high-dimensional manifolds. In promoter DNA design and protein backbone generation settings, RMF achieves comparable sample quality to prior methods while requiring up to 10× fewer function evaluations. Finally, we show that few-step flow maps enable efficient reward-guided design through reward look-ahead, where terminal states can be predicted from intermediate steps at minimal additional cost. RMF ICML 2026 Progressive Multi-Agent Reasoning for Biological Perturbation Prediction · Hyomin Kim, Sang-Yeon Hwang, Jaechang Lim, Yinhua Piao, Yunhak Oh, Woo Youn Kim, Chanyoung Park, Sungsoo Ahn, Junhyeok Jeon · arXiv 2026 · Predicting gene regulation responses to biological perturbations requires reasoning about underlying biological causalities. While large language models (LLMs) show promise for such tasks, they are often overwhelmed by the entangled nature of high-dimensional perturbation results. Moreover, recent works have primarily focused on genetic perturbations in single-cell experiments, leaving bulk-cell chemical perturbations, which is central to drug discovery, largely unexplored. Motivated by this, we present LINCSQA, a novel benchmark for predicting target gene regulation under complex chemical perturbations in bulk-cell environments. We further propose PBio-Agent, a multi-agent framework that integrates difficulty-aware task sequencing with iterative knowledge refinement. Our key insight is that genes affected by the same perturbation share causal structure, allowing confidently predicted genes to contextualize more challenging cases. The framework employs specialized agents enriched with biological knowledge graphs, while a synthesis agent integrates outputs and specialized judges ensure logical coherence. PBio-Agent outperforms existing baselines on both LINCSQA and PerturbQA, enabling even smaller models to predict and explain complex biological processes without additional training. PBio-Agent arXiv 2026 AtomMOF: All-Atom Flow Matching for MOF-Adsorbate Structure Prediction · Nayoung Kim, Honghui Kim, Sihyun Yu, Minkyu Kim, Seongsu Kim, Sungsoo Ahn · NeurIPS 2026 · Deep generative models have shown promise for modeling metal-organic frameworks (MOFs), but existing approaches (1) rely on coarse-grained representations that assume fixed bond lengths and angles, and (2) neglect the MOF-adsorbate interactions, which are critical for downstream applications. We introduce AtomMOF, a scalable flow-based model built on an all-atom Diffusion Transformer that maps 2D molecular graphs of building blocks and adsorbates directly to equilibrium 3D structures without imposing structural constraints. We further present scaling laws for porous crystal generation, indicating predictable performance gains with increased model capacity, and introduce Feynman-Kac steering guided by machine-learned interatomic potentials to improve geometric validity and sampling stability. On the (MOF-only) BW dataset, AtomMOF increases the match rate by 35.00% and reduces RMSD by 32.64%. On the ODAC25 dataset (MOF-adsorbate), AtomMOF is substantially more sample-efficient than grand canonical Monte Carlo in recovering adsorption configurations and can identify candidates with lower adsorption energies than the reference dataset. Code is available at https://github.com/nayoung10/AtomMOF . AtomMOF NeurIPS 2026 CatFlow: Co-generation of Slab-Adsorbate Systems via Flow Matching · Minkyu Kim, Nayoung Kim, Honghui Kim, Sungsoo Ahn · ICML 2026 · Discovering heterogeneous catalysts tailored for specific reaction intermediates remains a fundamental bottleneck in materials science. While traditional trial-and-error methods and recent generative models have shown promise, they struggle to capture the intrinsic coupling between surface geometry and adsorbate interactions. To address this limitation, we propose CatFlow, a flow matching-based framework for de novo design and structure prediction of heterogeneous catalysts. Our model operates on a primitive cell-based factorized representation of the slab-adsorbate complex, reducing the number of learnable variables by an average of 9.2x while explicitly encoding the surface orientation of the slab-adsorbate interface. Experiments on the Open Catalyst 2020 dataset demonstrate that CatFlow significantly improves the structural fidelity of generated catalysts compared to autoregressive and sequential baselines. Further experiments show that the generated structures accurately capture the adsorption energy distributions of physically plausible interfaces and lie closer to thermodynamic local minima. CatFlow ICML 2026 INDIBATOR: Diverse and Fact-Grounded Individuality for Multi-Agent Debate in Molecular Discovery · Yunhui Jang, Seonghyun Park, Jaehyung Kim, Sungsoo Ahn · arXiv 2026 · Multi-agent systems have emerged as a powerful paradigm for automating scientific discovery. To differentiate agent behavior in the multi-agent system, current frameworks typically assign generic role-based personas such as ''reviewer'' or ''writer'' or rely on coarse grained keyword-based personas. While functional, this approach oversimplifies how human scientists operate, whose contributions are shaped by their unique research trajectories. In response, we propose INDIBATOR, a framework for molecular discovery that grounds agents in individualized scientist profiles constructed from two modalities: publication history for literature-derived knowledge and molecular history for structural priors. These agents engage in multi-turn debate through proposal, critique, and voting phases. Our evaluation demonstrates that these fine-grained individuality-grounded agents consistently outperform systems relying on coarse-grained personas, achieving competitive or state-of-the-art performance. These results validate that capturing the ``scientific DNA'' of individual agents is essential for high-quality discovery. INDIBATOR arXiv 2026 A Systematic Evaluation of Co-folding Model Representations for Small-Molecule Learning · Hyosoon Jang, Hyunjin Seo, Honghui Kim, Taewon Kim, Yunhui Jang, Seonghyun Park, Sungsoo Ahn · NeurIPS 2026 · Small-molecule foundation models are typically pretrained on standalone molecular data, unlike vision and language models that often benefit from cross-modal or relational supervision. Protein-ligand co-folding provides a molecular analogue of such supervision by exposing models to atom-level ligand-protein interactions, raising the question of whether co-folding models can yield strong small-molecule representations. We study this question using Boltz2, a modern co-folding model, by transferring its atom-level ligand representations to standalone small-molecule tasks. Through systematic probing and distillation, we show that Boltz2 representations match or outperform existing models on the ADMET benchmark, accelerate molecular generative modeling, and improve sample efficiency in structure-guided ligand optimization. We further find that Boltz2 representations are complementary to those learned from conventional standalone molecular supervision, including 3D conformers, bioassay labels, and quantum-chemical properties. Finally, we extend representation alignment to reinforcement learning, showing that dense representation-level supervision can complement scalar rewards in molecular discovery. These results identify protein-ligand co-folding as a promising pretraining paradigm for small-molecule representation learning and position Boltz2 as a strong, off-the-shelf molecular foundation model. Co-folding reps NeurIPS 2026 DNACHUNKER: Learnable Tokenization for DNA Language Models · Taewon Kim, Jihwan Shin, Hyomin Kim, Youngmok Jung, Jonghoon Lee, Won-Chul Lee, Sungsoo Ahn†, Insu Han† · ICML 2026 · DNA language models are increasingly used to represent genomic sequence, yet their effectiveness depends critically on how raw nucleotides are converted into model inputs. Unlike natural language, DNA offers no canonical boundaries, making fixed tokenizations a brittle design choice under shifts, indels, and local repeats. We introduce DNAChunker, a masked DNA language model that incorporates a learnable adaptive segmentation module to produce context-dependent, variable-length units. Building on a dynamic segmentation procedure, DNAChunker learns to allocate finer granularity to functionally enriched regions while compressing repetitive or redundant sequence. We pretrain DNAChunker on the human reference genome and evaluate it across five benchmarks, where it consistently improves over strong fixed-tokenization baselines. Further analyses and ablations indicate that unlike fixed tokenizations, segmentation is learned in a biologically-informed, mutation-resilient manner. DNAChunker ICML 2026 Learning Collective Variables from BioEmu with Time-Lagged Generation · Seonghyun Park, Kiyoung Seong, Soojung Yang, Rafael Gómez-Bombarelli, Sungsoo Ahn · ICLR 2026 · Molecular dynamics is crucial for understanding molecular systems but its applicability is often limited by the vast timescales of rare events like protein folding. Enhanced sampling techniques overcome this by accelerating the simulation along key reaction pathways, which are defined by collective variables (CVs). However, identifying effective CVs that capture the slow, macroscopic dynamics of a system remains a major bottleneck. This work proposes a novel framework coined BioEmu-CV that learns these essential CVs automatically from BioEmu, a recently proposed foundation model for generating protein equilibrium samples. In particular, we re-purpose BioEmu to learn time-lagged generation conditioned on the learned CV, i.e., predict the distribution of molecular states after a certain amount of time. This training process promotes the CV to encode only the slow, long-term information while disregarding fast, random fluctuations. We validate our learned CV on fast-folding proteins with two key applications: (1) estimating free energy differences using on-the-fly probability enhanced sampling and (2) sampling transition paths with steered molecular dynamics. Our empirical study also serves as a new systematic and comprehensive benchmark for MLCVs on fast-folding proteins larger than Alanine Dipeptide. BioEmu-CV ICLR 2026 MT-Mol: Multi Agent System with Tool-based Reasoning for Molecular Optimization · Hyomin Kim, Yunhui Jang, Sungsoo Ahn · EMNLP 2025 · Large language models (LLMs) have large potential for molecular optimization, as they can gather external chemistry tools and enable collaborative interactions to iteratively refine molecular candidates. However, this potential remains underexplored, particularly in the context of structured reasoning, interpretability, and comprehensive tool-grounded molecular optimization. To address this gap, we introduce MT-Mol, a multi-agent framework for molecular optimization that leverages tool-guided reasoning and role-specialized LLM agents. Our system incorporates comprehensive RDKit tools, categorized into five distinct domains: structural descriptors, electronic and topological features, fragment-based functional groups, molecular representations, and miscellaneous chemical properties. Each category is managed by an expert analyst agent, responsible for extracting task-relevant tools and enabling interpretable, chemically grounded feedback. MT-Mol produces molecules with tool-aligned and stepwise reasoning through the interaction between the analyst agents, a molecule-generating scientist, a reasoning-output verifier, and a reviewer agent. As a result, we show that our framework shows the state-of-the-art performance of the PMO-1K benchmark on 17 out of 23 tasks. MT-Mol EMNLP 2025 On Scalable and Efficient Training of Diffusion Samplers · Minkyu Kim, Kiyoung Seong, Dongyeop Woo, Sungsoo Ahn, Minsu Kim · NeurIPS 2025 · We address the challenge of training diffusion models to sample from unnormalized energy distributions in the absence of data, the so-called diffusion samplers. Although these approaches have shown promise, they struggle to scale in more demanding scenarios where energy evaluations are expensive and the sampling space is high-dimensional. To address this limitation, we propose a scalable and sample-efficient framework that properly harmonizes the powerful classical sampling method and the diffusion sampler. Specifically, we utilize Monte Carlo Markov chain (MCMC) samplers with a novelty-based auxiliary energy as a Searcher to collect off-policy samples, using an auxiliary energy function to compensate for exploring modes the diffusion sampler rarely visits. These off-policy samples are then combined with on-policy data to train the diffusion sampler, thereby expanding its coverage of the energy landscape. Furthermore, we identify primacy bias, i.e., the preference of samplers for early experience during training, as the main cause of mode collapse during training, and introduce a periodic re-initialization trick to resolve this issue. Our method significantly improves sample efficiency on standard benchmarks for diffusion samplers and also excels at higher-dimensional problems and real-world molecular conformer generation. SGDS NeurIPS 2025 Energy-based Generator Matching: A Neural Sampler for General State Space · Dongyeop Woo, Minsu Kim, Minkyu Kim, Kiyoung Seong, Sungsoo Ahn · NeurIPS 2025 · We propose Energy-based generator matching (EGM), a modality-agnostic approach to train generative models from energy functions in the absence of data. Extending the recently proposed generator matching, EGM enables training of arbitrary continuous-time Markov processes, e.g., diffusion, flow, and jump, and can generate data from continuous, discrete, and a mixture of two modalities. To this end, we propose estimating the generator matching loss using self-normalized importance sampling with an additional bootstrapping trick to reduce variance in the importance weight. We validate EGM on both discrete and multimodal tasks up to 100 and 20 dimensions, respectively. EGM NeurIPS 2025 High-order Equivariant Flow Matching for Density Functional Theory Hamiltonian Prediction · Seongsu Kim, Nayoung Kim, Dongwoo Kim, Sungsoo Ahn · NeurIPS 2025 · Density functional theory (DFT) is a fundamental method for simulating quantum chemical properties, but it remains expensive due to the iterative self-consistent field (SCF) process required to solve the Kohn-Sham equations. Recently, deep learning methods are gaining attention as a way to bypass this step by directly predicting the Hamiltonian. However, they rely on deterministic regression and do not consider the highly structured nature of Hamiltonians. In this work, we propose QHFlow, a high-order equivariant flow matching framework that generates Hamiltonian matrices conditioned on molecular geometry. Flow matching models continuous-time trajectories between simple priors and complex targets, learning the structured distributions over Hamiltonians instead of direct regression. To further incorporate symmetry, we use a neural architecture that predicts SE(3)-equivariant vector fields, improving accuracy and generalization across diverse geometries. To further enhance physical fidelity, we additionally introduce a fine-tuning scheme to align predicted orbital energies with the target. QHFlow achieves state-of-the-art performance, reducing Hamiltonian error by 71% on MD17 and 53% on QH9. Moreover, we further show that QHFlow accelerates the DFT process without trading off the solution quality when initializing SCF iterations with the predicted Hamiltonian, significantly reducing the number of iterations and runtime. QHFlow NeurIPS 2025 Self-Training Large Language Models with Confident Reasoning · Hyosoon Jang, Yunhui Jang, Sungjae Lee, Jungseul Ok, Sungsoo Ahn · EMNLP 2025 · Large language models (LLMs) have shown impressive performance by generating reasoning paths before final answers, but learning such a reasoning path requires costly human supervision. To address this issue, recent studies have explored self-training methods that improve reasoning capabilities using pseudo-labels generated by the LLMs themselves. Among these, confidence-based self-training fine-tunes LLMs to prefer reasoning paths with high-confidence answers, where confidence is estimated via majority voting. However, such methods exclusively focus on the quality of the final answer and may ignore the quality of the reasoning paths, as even an incorrect reasoning path leads to a correct answer by chance. Instead, we advocate the use of reasoning-level confidence to identify high-quality reasoning paths for self-training, supported by our empirical observations. We then propose a new self-training method, CORE-PO, that fine-tunes LLMs to prefer high-COnfidence REasoning paths through Policy Optimization. Our experiments show that CORE-PO improves the accuracy of outputs on four in-distribution and two out-of-distribution benchmarks, compared to existing self-training methods. CORE-PO EMNLP 2025 Flexible MOF Generation with Torsion-Aware Flow Matching · Nayoung Kim, Seongsu Kim, Sungsoo Ahn · NeurIPS 2025 · Designing metal-organic frameworks (MOFs) with novel chemistries is a longstanding challenge due to their large combinatorial space and complex 3D arrangements of the building blocks. While recent deep generative models have enabled scalable MOF generation, they assume (1) a fixed set of building blocks and (2) known local 3D coordinates of building blocks. However, this limits their ability to (1) design novel MOFs and (2) generate the structure using novel building blocks. We propose a two-stage MOF generation framework that overcomes these limitations by modeling both chemical and geometric degrees of freedom. First, we train an SMILES-based autoregressive model to generate metal and organic building blocks, paired with a cheminformatics toolkit for 3D structure initialization. Second, we introduce a flow matching model that predicts translations, rotations, and torsional angles to assemble the blocks into valid 3D frameworks. Our experiments demonstrate improved reconstruction accuracy, the generation of valid, novel, and unique MOFs, and the ability to create novel building blocks. Our code is available at https://github.com/nayoung10/MOFFlow-2 . MOFFlow-2 NeurIPS 2025 Learning Flexible Forward Trajectories for Masked Molecular Diffusion · Hyunjin Seo*, Taewon Kim*, Sihyun Yu, Sungsoo Ahn · ICLR 2026 · Masked diffusion models (MDMs) have achieved notable progress in modeling discrete data, while their potential in molecular generation remains underexplored. In this work, we explore their potential and introduce the surprising result that naively applying standards MDMs severely degrades the performance. We identify the critical cause of this issue as a state-clashing problem-where the forward diffusion of distinct molecules collapse into a common state, resulting in a mixture of reconstruction targets that cannot be learned using typical reverse diffusion process with unimodal predictions. To mitigate this, we propose Masked Element-wise Learnable Diffusion (MELD) that orchestrates per-element corruption trajectories to avoid collision between distinct molecular graphs. This is achieved through a parameterized noise scheduling network that assigns distinct corruption rates to individual graph elements, i.e., atoms and bonds. Extensive experiments on diverse molecular benchmarks reveal that MELD markedly enhances overall generation quality compared to element-agnostic noise scheduling, increasing the chemical validity of vanilla MDMs on ZINC250K from 15% to 93%, Furthermore, it achieves state-of-the-art property alignment in conditional generation tasks. MELD ICLR 2026 Improving Chemical Understanding of LLMs via SMILES Parsing · Yunhui Jang, Jaehyung Kim, Sungsoo Ahn · EMNLP 2025 · Large language models (LLMs) are increasingly recognized as powerful tools for scientific discovery, particularly in molecular science. A fundamental requirement for these models is the ability to accurately understand molecular structures, commonly encoded in the SMILES representation. However, current LLMs struggle to interpret SMILES, even failing to carry out basic tasks such as counting molecular rings. To address this limitation, we introduce CLEANMOL, a novel framework that formulates SMILES parsing into a suite of clean and deterministic tasks explicitly designed to promote graph-level molecular comprehension. These tasks span from subgraph matching to global graph matching, providing structured supervision aligned with molecular structural properties. We construct a molecular pretraining dataset with adaptive difficulty scoring and pre-train open-source LLMs on these tasks. Our results show that CLEANMOL not only enhances structural comprehension but also achieves the best or competes with the baseline on the Mol-Instructions benchmark. CleanMol EMNLP 2025 Latent Veracity Inference for Identifying Errors in Stepwise Reasoning · Minsu Kim, Jean-Pierre R. Falet, Oliver Ethan Richardson, Xiaoyin Chen, Moksh Jain, Sungjin Ahn, Sungsoo Ahn, Yoshua Bengio · ICLR 2026 · Chain-of-Thought (CoT) reasoning has advanced the capabilities and transparency of language models (LMs); however, reasoning chains can contain inaccurate statements that reduce performance and trustworthiness. To address this, we propose to augment each reasoning step in a CoT with a latent veracity (or correctness) variable. To efficiently explore this expanded space, we introduce Veracity Search (VS), a discrete search algorithm over veracity assignments. It performs otherwise intractable inference in the posterior distribution over latent veracity values by leveraging the LM's joint likelihood over veracity and the final answer as a proxy reward. This efficient inference-time verification method facilitates supervised fine-tuning of an Amortized Veracity Inference (AVI) machine by providing pseudo-labels for veracity. AVI generalizes VS, enabling accurate zero-shot veracity inference in novel contexts. Empirical results demonstrate that VS reliably identifies errors in logical (ProntoQA), mathematical (GSM8K), and commonsense (CommonsenseQA) reasoning benchmarks, with AVI achieving comparable zero-shot accuracy. Finally, we demonstrate the utility of latent veracity inference for providing feedback during self-correction and self-improvement. VS / AVI ICLR 2026 Enhancing LLM Agent Safety via Causal Influence Prompting · Dongyoon Hahm, Woogyeol Jin, June Suk Choi, Sungsoo Ahn, Kimin Lee · ACL 2025 · As autonomous agents powered by large language models (LLMs) continue to demonstrate potential across various assistive tasks, ensuring their safe and reliable behavior is crucial for preventing unintended consequences. In this work, we introduce CIP, a novel technique that leverages causal influence diagrams (CIDs) to identify and mitigate risks arising from agent decision-making. CIDs provide a structured representation of cause-and-effect relationships, enabling agents to anticipate harmful outcomes and make safer decisions. Our approach consists of three key steps: (1) initializing a CID based on task specifications to outline the decision-making process, (2) guiding agent interactions with the environment using the CID, and (3) iteratively refining the CID based on observed behaviors and outcomes. Experimental results demonstrate that our method effectively enhances safety in both code execution and mobile device control tasks. CIP ACL 2025 Structural Reasoning Improves Molecular Understanding of LLM · Yunhui Jang, Jaehyung Kim, Sungsoo Ahn · ACL 2025 · Recently, large language models (LLMs) have shown significant progress, approaching human perception levels. In this work, we demonstrate that despite these advances, LLMs still struggle to reason using molecular structural information. This gap is critical because many molecular properties, including functional groups, depend heavily on such structural details. To address this limitation, we propose an approach that sketches molecular structures for reasoning. Specifically, we introduce Molecular Structural Reasoning (MSR) framework to enhance the understanding of LLMs by explicitly incorporating the key structural features. We present two frameworks for scenarios where the target molecule is known or unknown. We verify that our MSR improves molecular understanding through extensive experiments. MSR ACL 2025 MOFFlow: Flow Matching for Structure Prediction of Metal-Organic Frameworks · Nayoung Kim, Seongsu Kim, Minsu Kim, Jinkyoo Park, Sungsoo Ahn · ICLR 2025 · Metal-organic frameworks (MOFs) are a class of crystalline materials with promising applications in many areas such as carbon capture and drug delivery. In this work, we introduce MOFFlow, the first deep generative model tailored for MOF structure prediction. Existing approaches, including ab initio calculations and even deep generative models, struggle with the complexity of MOF structures due to the large number of atoms in the unit cells. To address this limitation, we propose a novel Riemannian flow matching framework that reduces the dimensionality of the problem by treating the metal nodes and organic linkers as rigid bodies, capitalizing on the inherent modularity of MOFs. By operating in the SE(3) space, MOFFlow effectively captures the roto-translational dynamics of these rigid components in a scalable way. Our experiment demonstrates that MOFFlow accurately predicts MOF structures containing several hundred atoms, significantly outperforming conventional methods and state-of-the-art machine learning baselines while being much faster. MOFFlow ICLR 2025 Generative Flows on Synthetic Pathway for Drug Design · Seonghwan Seo, Minsu Kim, Tony Shen, Martin Ester, Jinkyoo Park, Sungsoo Ahn, Woo Youn Kim · ICLR 2025 · Generative models in drug discovery have recently gained attention as efficient alternatives to brute-force virtual screening. However, most existing models do not account for synthesizability, limiting their practical use in real-world scenarios. In this paper, we propose RxnFlow, which sequentially assembles molecules using predefined molecular building blocks and chemical reaction templates to constrain the synthetic chemical pathway. We then train on this sequential generating process with the objective of generative flow networks (GFlowNets) to generate both highly rewarded and diverse molecules. To mitigate the large action space of synthetic pathways in GFlowNets, we implement a novel action space subsampling method. This enables RxnFlow to learn generative flows over extensive action spaces comprising combinations of 1.2 million building blocks and 71 reaction templates without significant computational overhead. Additionally, RxnFlow can employ modified or expanded action spaces for generation without retraining, allowing for the introduction of additional objectives or the incorporation of newly discovered building blocks. We experimentally demonstrate that RxnFlow outperforms existing reaction-based and fragment-based models in pocket-specific optimization across various target pockets. Furthermore, RxnFlow achieves state-of-the-art performance on CrossDocked2020 for pocket-conditional generation, with an average Vina score of -8.85 kcal/mol and 34.8% synthesizability. RxnFlow ICLR 2025 ReBind: Enhancing Ground-state Molecular Conformation Prediction via Force-Based Graph Rewiring · Taewon Kim*, Hyunjin Seo*, Sungsoo Ahn, Eunho Yang · ICLR 2025 · Predicting the ground-state 3D molecular conformations from 2D molecular graphs is critical in computational chemistry due to its profound impact on molecular properties. Deep learning (DL) approaches have recently emerged as promising alternatives to computationally-heavy classical methods such as density functional theory (DFT). However, we discover that existing DL methods inadequately model inter-atomic forces, particularly for non-bonded atomic pairs, due to their naive usage of bonds and pairwise distances. Consequently, significant prediction errors occur for atoms with low degree (i.e., low coordination numbers) whose conformations are primarily influenced by non-bonded interactions. To address this, we propose REBIND, a novel framework that rewires molecular graphs by adding edges based on the Lennard-Jones potential to capture non-bonded interactions for low-degree atoms. Experimental results demonstrate that REBIND significantly outperforms state-of-the-art methods across various molecular sizes, achieving up to a 20% reduction in prediction error. ReBind ICLR 2025 Can LLMs Generate Diverse Molecules? Towards Alignment with Structural Diversity · Hyosoon Jang, Yunhui Jang, Jaehyung Kim, Sungsoo Ahn · arXiv 2024 · Recent advancements in large language models (LLMs) have demonstrated impressive performance in molecular generation, which offers potential to accelerate drug discovery. However, the current LLMs overlook a critical requirement for drug discovery: proposing a diverse set of molecules. This diversity is essential for improving the chances of finding a viable drug, as it provides alternative molecules that may succeed where others fail in real-world validations. Nevertheless, the LLMs often output structurally similar molecules. While decoding schemes like diverse beam search may enhance textual diversity, this often does not align with molecular structural diversity. In response, we propose a new method for fine-tuning molecular generative LLMs to autoregressively generate a set of structurally diverse molecules, where each molecule is generated by conditioning on the previously generated molecules. Our approach consists of two stages: (1) supervised fine-tuning to adapt LLMs to autoregressively generate molecules in a sequence and (2) reinforcement learning to maximize structural diversity within the generated molecules. Our experiments show that the proposed approach enables LLMs to generate diverse molecules better than existing approaches for diverse sequence generation. Div-SFT+RL arXiv 2024 Adaptive Teachers for Amortized Samplers · Minsu Kim*, Sanghyeok Choi*, Taeyoung Yun, Emmanuel Bengio, Leo Feng, Jarrid Rector-Brooks, Sungsoo Ahn, Jinkyoo Park, Esmeralda S. Whitammer, Yoshua Bengio · ICLR 2025 · Amortized inference is the task of training a parametric model, such as a neural network, to approximate a distribution with a given unnormalized density where exact sampling is intractable. When sampling is implemented as a sequential decision-making process, reinforcement learning (RL) methods, such as generative flow networks, can be used to train the sampling policy. Off-policy RL training facilitates the discovery of diverse, high-reward candidates, but existing methods still face challenges in efficient exploration. We propose to use an adaptive training distribution (the teacher) to guide the training of the primary amortized sampler (the student). The teacher, an auxiliary behavior model, is trained to sample high-loss regions of the student and can generalize across unexplored modes, thereby enhancing mode coverage by providing an efficient training curriculum. We validate the effectiveness of this approach in a synthetic environment designed to present an exploration challenge, two diffusion-based sampling tasks, and four biochemical discovery tasks demonstrating its ability to improve sample efficiency and mode coverage. Source code is available at https://github.com/alstn12088/adaptive-teacher . Adaptive Teachers ICLR 2025 Iterated Energy-based Flow Matching for Sampling from Boltzmann Densities · Dongyeop Woo, Sungsoo Ahn · arXiv 2024 · In this work, we consider the problem of training a generator from evaluations of energy functions or unnormalized densities. This is a fundamental problem in probabilistic inference, which is crucial for scientific applications such as learning the 3D coordinate distribution of a molecule. To solve this problem, we propose iterated energy-based flow matching (iEFM), the first off-policy approach to train continuous normalizing flow (CNF) models from unnormalized densities. We introduce the simulation-free energy-based flow matching objective, which trains the model to predict the Monte Carlo estimation of the marginal vector field constructed from known energy functions. Our framework is general and can be extended to variance-exploding (VE) and optimal transport (OT) conditional probability paths. We evaluate iEFM on a two-dimensional Gaussian mixture model (GMM) and an eight-dimensional four-particle double-well potential (DW-4) energy function. Our results demonstrate that iEFM outperforms existing methods, showcasing its potential for efficient and scalable probabilistic modeling in complex high-dimensional systems. iEFM arXiv 2024 RL4CO: an Extensive Reinforcement Learning for Combinatorial Optimization Benchmark · Federico Berto, Chuanbo Hua, Junyoung Park, Laurin Luttmann, Yining Ma, Fanchen Bu, Jiarui Wang, Haoran Ye, Minsu Kim, Sanghyeok Choi, Nayeli Gast Zepeda, André Hottung, Jianan Zhou, Jieyi Bi, Yu Hu, Fei Liu, Hyeonah Kim, Jiwoo Son, Haeyeon Kim, Davide Angioni, Wouter Kool, Zhiguang Cao, Qingfu Zhang, Joungho Kim, Jie Zhang, Kijung Shin, Cathy Wu, Sungsoo Ahn, Guojie Song, Changhyun Kwon, Kevin Tierney, Lin Xie, Jinkyoo Park · KDD 2025 · Combinatorial optimization (CO) is fundamental to several real-world applications, from logistics and scheduling to hardware design and resource allocation. Deep reinforcement learning (RL) has recently shown significant benefits in solving CO problems, reducing reliance on domain expertise and improving computational efficiency. However, the absence of a unified benchmarking framework leads to inconsistent evaluations, limits reproducibility, and increases engineering overhead, raising barriers to adoption for new researchers. To address these challenges, we introduce RL4CO, a unified and extensive benchmark with in-depth library coverage of 27 CO problem environments and 23 state-of-the-art baselines. Built on efficient software libraries and best practices in implementation, RL4CO features modularized implementation and flexible configurations of diverse environments, policy architectures, RL algorithms, and utilities with extensive documentation. RL4CO helps researchers build on existing successes while exploring and developing their own designs, facilitating the entire research process by decoupling science from heavy engineering. We finally provide extensive benchmark studies to inspire new insights and future work. RL4CO has already attracted numerous researchers in the community and is open-sourced at https://github.com/ai4co/rl4co . RL4CO KDD 2025 Holistic Molecular Representation Learning via Multi-view Fragmentation · Seojin Kim, Jaehyun Nam, Junsu Kim, Hankook Lee, Sungsoo Ahn, Jinwoo Shin · TMLR 2024 · Learning chemically meaningful representations from unlabeled molecules plays a vital role in AI-based drug design and discovery. In response to this, several self-supervised learning methods have been developed, focusing either on global (e.g., graph-level) or local (e.g., motif-level) information of molecular graphs. However, it is still unclear which approach is more effective for learning better molecular representations. In this paper, we propose a novel holistic self-supervised molecular representation learning framework that effectively learns both global and local molecular information. Our key idea is to utilize fragmentation, which decomposes a molecule into a set of chemically meaningful fragments (e.g., functional groups), to associate a global graph structure to a set of local substructures, thereby preserving chemical properties and learn both information via contrastive learning between them. Additionally, we also consider the 3D geometry of molecules as another view for contrastive learning. We demonstrate that our framework outperforms prior molecular representation learning methods across various molecular property prediction tasks. HoliMol TMLR 2024 Transition Path Sampling with Improved Off-Policy Training of Diffusion Path Samplers · Kiyoung Seong, Seonghyun Park, Seonghwan Kim, Woo Youn Kim, Sungsoo Ahn · ICLR 2025 · Understanding transition pathways between two meta-stable states of a molecular system is crucial to advance drug discovery and material design. However, unbiased molecular dynamics (MD) simulations are computationally infeasible because of the high energy barriers that separate these states. Although recent machine learning techniques are proposed to sample rare events, they are often limited to simple systems and rely on collective variables (CVs) derived from costly domain expertise. In this paper, we introduce a novel approach that trains diffusion path samplers (DPS) to address the transition path sampling (TPS) problem without requiring CVs. We reformulate the problem as an amortized sampling from the transition path distribution by minimizing the log-variance divergence between the path distribution induced by DPS and the transition path distribution. Based on the log-variance divergence, we propose learnable control variates to reduce the variance of gradient estimators and the off-policy training objective with replay buffers and simulated annealing techniques to improve sample efficiency and diversity. We also propose a scale-based equivariant parameterization of the bias forces to ensure scalability for large systems. We extensively evaluate our approach, termed TPS-DPS, on a synthetic system, small peptide, and challenging fast-folding proteins, demonstrating that it produces more realistic and diverse transition pathways than existing baselines. TPS-DPS ICLR 2025 Decoupled Sequence and Structure Generation for Realistic Antibody Design · Nayoung Kim, Minsu Kim, Sungsoo Ahn, Jinkyoo Park · TMLR 2024 · Recently, deep learning has made rapid progress in antibody design, which plays a key role in the advancement of therapeutics. A dominant paradigm is to train a model to jointly generate the antibody sequence and the structure as a candidate. However, the joint generation requires the model to generate both the discrete amino acid categories and the continuous 3D coordinates; this limits the space of possible architectures and may lead to suboptimal performance. In response, we propose an antibody sequence-structure decoupling (ASSD) framework, which separates sequence generation and structure prediction. Although our approach is simple, our idea allows the use of powerful neural architectures and demonstrates notable performance improvements. We also find that the widely used non-autoregressive generators promote sequences with overly repeating tokens. Such sequences are both out-of-distribution and prone to undesirable developability properties that can trigger harmful immune responses in patients. To resolve this, we introduce a composition-based objective that allows an efficient trade-off between high performance and low token repetition. ASSD shows improved performance in various antibody design experiments, while the composition-based objective successfully mitigates token repetition of non-autoregressive models. ASSD TMLR 2024 Pessimistic Backward Policy for GFlowNets · Hyosoon Jang, Yunhui Jang, Minsu Kim, Jinkyoo Park, Sungsoo Ahn · NeurIPS 2024 · This paper studies Generative Flow Networks (GFlowNets), which learn to sample objects proportionally to a given reward function through the trajectory of state transitions. In this work, we observe that GFlowNets tend to under-exploit the high-reward objects due to training on insufficient number of trajectories, which may lead to a large gap between the estimated flow and the (known) reward value. In response to this challenge, we propose a pessimistic backward policy for GFlowNets (PBP-GFN), which maximizes the observed flow to align closely with the true reward for the object. We extensively evaluate PBP-GFN across eight benchmarks, including hyper-grid environment, bag generation, structured set generation, molecular generation, and four RNA sequence generation tasks. In particular, PBP-GFN enhances the discovery of high-reward objects, maintains the diversity of the objects, and consistently outperforms existing methods. PBP-GFN NeurIPS 2024 Tackling Complex Conditions in Unsupervised Combinatorial Optimization · Fanchen Bu, Hyeonsoo Jo, Soo Yong Lee, Sungsoo Ahn, Kijung Shin · ICML 2024 · Combinatorial optimization (CO) is naturally discrete, making machine learning based on differentiable optimization inapplicable. Karalias & Loukas (2020) adapted the probabilistic method to incorporate CO into differentiable optimization. Their work ignited the research on unsupervised learning for CO, composed of two main components: probabilistic objectives and derandomization. However, each component confronts unique challenges. First, deriving objectives under various conditions (e.g., cardinality constraints and minimum) is nontrivial. Second, the derandomization process is underexplored, and the existing derandomization methods are either random sampling or naive rounding. In this work, we aim to tackle prevalent (i.e., commonly involved) conditions in unsupervised CO. First, we concretize the targets for objective construction and derandomization with theoretical justification. Then, for various conditions commonly involved in different CO problems, we derive nontrivial objectives and derandomization to meet the targets. Finally, we apply the derivations to various CO problems. Via extensive experiments on synthetic and real-world graphs, we validate the correctness of our derivations and show our empirical superiority w.r.t. both optimization quality and speed. UCom2 ICML 2024 Improving Robustness to Multiple Spurious Correlations by Multi-Objective Optimization · Nayeong Kim, Juwon Kang, Sungsoo Ahn, Jungseul Ok, Suha Kwak · ICML 2024 · We study the problem of training an unbiased and accurate model given a dataset with multiple biases. This problem is challenging since the multiple biases cause multiple undesirable shortcuts during training, and even worse, mitigating one may exacerbate the other. We propose a novel training method to tackle this challenge. Our method first groups training data so that different groups induce different shortcuts, and then optimizes a linear combination of group-wise losses while adjusting their weights dynamically to alleviate conflicts between the groups in performance; this approach, rooted in the multi-objective optimization theory, encourages to achieve the minimax Pareto solution. We also present a new benchmark with multiple biases, dubbed MultiCelebA, for evaluating debiased training methods under realistic and challenging scenarios. Our method achieved the best on three datasets with multiple biases, and also showed superior performance on conventional single-bias datasets. Multi-bias learning ICML 2024 Breadth-First Exploration in Adaptive Grid-based Reinforcement Learning · Youngsik Yoon, Gangbok Lee, Sungsoo Ahn, Jungseul Ok · ICML 2024 · Graph-based planners have gained significant attention for goal-conditioned reinforcement learning (RL), where they construct a graph consisting of confident transitions between subgoals as edges and run shortest path algorithms to exploit the confident edges. Meanwhile, identifying and avoiding unattainable transitions are also crucial yet overlooked by the previous graph-based planners, leading to wasting an excessive number of attempts at unattainable subgoals. To address this oversight, we propose a graph construction method that efficiently manages all the achieved and unattained subgoals on a grid graph adaptively discretizing the goal space. This enables a breadth-first exploration strategy, grounded in the local adaptive grid refinement, that prioritizes broad probing of subgoals on a coarse grid over meticulous one on a dense grid. We conducted a theoretical analysis and demonstrated the effectiveness of our approach through empirical evidence, showing that only BEAG succeeds in complex environments under the proposed fixed-goal setting. BEAG ICML 2024 Hybrid Neural Representation for Spherical Data · Hyomin Kim, Yunhui Jang, Jaeho Lee, Sungsoo Ahn · ICML 2024 · In this paper, we study hybrid neural representations for spherical data, a domain of increasing relevance in scientific research. In particular, our work focuses on weather and climate data as well as comic microwave background (CMB) data. Although previous studies have delved into coordinate-based neural representations for spherical signals, they often fail to capture the intricate details of highly nonlinear signals. To address this limitation, we introduce a novel approach named Hybrid Neural Representations for Spherical data (HNeR-S). Our main idea is to use spherical feature-grids to obtain positional features which are combined with a multilayer perception to predict the target signal. We consider feature-grids with equirectangular and hierarchical equal area isolatitude pixelization structures that align with weather data and CMB data, respectively. We extensively verify the effectiveness of our HNeR-S for regression, super-resolution, temporal interpolation, and compression tasks. HNeR-S ICML 2024 Gaussian Plane-Wave Neural Operator for Electron Density Estimation · Seongsu Kim, Sungsoo Ahn · ICML 2024 · This work studies machine learning for electron density prediction, which is fundamental for understanding chemical systems and density functional theory (DFT) simulations. To this end, we introduce the Gaussian plane-wave neural operator (GPWNO), which operates in the infinite-dimensional functional space using the plane-wave and Gaussian-type orbital bases, widely recognized in the context of DFT. In particular, both high- and low-frequency components of the density can be effectively represented due to the complementary nature of the two bases. Extensive experiments on QM9, MD, and material project datasets demonstrate GPWNO's superior performance over ten baselines. GPWNO ICML 2024 A Simple and Scalable Representation for Graph Generation · Yunhui Jang, Seul Lee, Sungsoo Ahn · ICLR 2024 · Recently, there has been a surge of interest in employing neural networks for graph generation, a fundamental statistical learning problem with critical applications like molecule design and community analysis. However, most approaches encounter significant limitations when generating large-scale graphs. This is due to their requirement to output the full adjacency matrices whose size grows quadratically with the number of nodes. In response to this challenge, we introduce a new, simple, and scalable graph representation named gap encoded edge list (GEEL) that has a small representation size that aligns with the number of edges. In addition, GEEL significantly reduces the vocabulary size by incorporating the gap encoding and bandwidth restriction schemes. GEEL can be autoregressively generated with the incorporation of node positional encoding, and we further extend GEEL to deal with attributed graphs by designing a new grammar. Our findings reveal that the adoption of this compact representation not only enhances scalability but also bolsters performance by simplifying the graph generation process. We conduct a comprehensive evaluation across ten non-attributed and two molecular graph generation tasks, demonstrating the effectiveness of GEEL. GEEL ICLR 2024 Non-backtracking Graph Neural Networks · Seonghyun Park*, Narae Ryu*, Gahee Kim, Dongyeop Woo, Se-Young Yun†, Sungsoo Ahn† · TMLR 2024 · The celebrated message-passing updates for graph neural networks allow representing large-scale graphs with local and computationally tractable updates. However, the updates suffer from backtracking, i.e., a message flowing through the same edge twice and revisiting the previously visited node. Since the number of message flows increases exponentially with the number of updates, the redundancy in local updates prevents the graph neural network from accurately recognizing a particular message flow relevant for downstream tasks. In this work, we propose to resolve such a redundancy issue via the non-backtracking graph neural network (NBA-GNN) that updates a message without incorporating the message from the previously visited node. We theoretically investigate how NBA-GNN alleviates the over-squashing of GNNs, and establish a connection between NBA-GNN and the impressive performance of non-backtracking updates for stochastic block model recovery. Furthermore, we empirically verify the effectiveness of our NBA-GNN on the long-range graph benchmark and transductive node classification problems. NBA-GNN TMLR 2024 Learning Energy Decompositions for Partial Inference in GFlowNets · Hyosoon Jang, Minsu Kim, Sungsoo Ahn · ICLR 2024 · This paper studies generative flow networks (GFlowNets) to sample objects from the Boltzmann energy distribution via a sequence of actions. In particular, we focus on improving GFlowNet with partial inference: training flow functions with the evaluation of the intermediate states or transitions. To this end, the recently developed forward-looking GFlowNet reparameterizes the flow functions based on evaluating the energy of intermediate states. However, such an evaluation of intermediate energies may (i) be too expensive or impossible to evaluate and (ii) even provide misleading training signals under large energy fluctuations along the sequence of actions. To resolve this issue, we propose learning energy decompositions for GFlowNets (LED-GFN). Our main idea is to (i) decompose the energy of an object into learnable potential functions defined on state transitions and (ii) reparameterize the flow functions using the potential functions. In particular, to produce informative local credits, we propose to regularize the potential to change smoothly over the sequence of actions. It is also noteworthy that training GFlowNet with our learned potential can preserve the optimal policy. We empirically verify the superiority of LED-GFN in five problems including the generation of unstructured and maximum independent sets, molecular graphs, and RNA sequences. LED-GFN ICLR 2024 Local Search GFlowNets · Minsu Kim, Taeyoung Yun, Emmanuel Bengio, Dinghuai Zhang, Yoshua Bengio, Sungsoo Ahn, Jinkyoo Park · ICLR 2024 · Generative Flow Networks (GFlowNets) are amortized sampling methods that learn a distribution over discrete objects proportional to their rewards. GFlowNets exhibit a remarkable ability to generate diverse samples, yet occasionally struggle to consistently produce samples with high rewards due to over-exploration on wide sample space. This paper proposes to train GFlowNets with local search, which focuses on exploiting high-rewarded sample space to resolve this issue. Our main idea is to explore the local neighborhood via backtracking and reconstruction guided by backward and forward policies, respectively. This allows biasing the samples toward high-reward solutions, which is not possible for a typical GFlowNet solution generation scheme, which uses the forward policy to generate the solution from scratch. Extensive experiments demonstrate a remarkable performance improvement in several biochemical tasks. Source code is available: https://github.com/dbsxodud-11/ls_gfn. LS-GFN ICLR 2024 Multi-resolution Spectral Coherence for Graph Generation with Score-based Diffusion · Hyuna Cho, Minjae Jeong, Sooyeon Jeon, Sungsoo Ahn, Won Hwa Kim · NeurIPS 2023 · Successful graph generation depends on the accurate estimation of the joint distribution of graph components such as nodes and edges from training data. While recent deep neural networks have demonstrated sampling of realistic graphs together with diffusion models, however, they still suffer from oversmoothing problems which are inherited from conventional graph convolution and thus high-frequency characteristics of nodes and edges become intractable. To overcome such issues and generate graphs with high fidelity, this paper introduces a novel approach that captures the dependency between nodes and edges at multiple resolutions in the spectral space. By modeling the joint distribution of node and edge signals in a shared graph wavelet space, together with a score-based diffusion model, we propose a Wavelet Graph Diffusion Model (Wave-GD) which lets us sample synthetic graphs with real-like frequency characteristics of nodes and edges. Experimental results on four representative benchmark datasets validate the superiority of the Wave-GD over existing approaches, highlighting its potential for a wide range of applications that involve graph data. Wave-GD NeurIPS 2023 Bootstrapped Training of Score-Conditioned Generator for Offline Design of Biological Sequences · Minsu Kim, Federico Berto, Sungsoo Ahn, Jinkyoo Park · NeurIPS 2023 · We study the problem of optimizing biological sequences, e.g., proteins, DNA, and RNA, to maximize a black-box score function that is only evaluated in an offline dataset. We propose a novel solution, bootstrapped training of score-conditioned generator (BootGen) algorithm. Our algorithm repeats a two-stage process. In the first stage, our algorithm trains the biological sequence generator with rank-based weights to enhance the accuracy of sequence generation based on high scores. The subsequent stage involves bootstrapping, which augments the training dataset with self-generated data labeled by a proxy score function. Our key idea is to align the score-based generation with a proxy score function, which distills the knowledge of the proxy score function to the generator. After training, we aggregate samples from multiple bootstrapped generators and proxies to produce a diverse design. Extensive experiments show that our method outperforms competitive baselines on biological sequential design tasks. We provide reproducible source code: https://github.com/kaist-silab/bootgen. BootGen NeurIPS 2023 EPIC: Graph Augmentation with Edit Path Interpolation via Learnable Cost · Jaeseung Heo, Seungbeom Lee, Sungsoo Ahn, Dongwoo Kim · IJCAI 2024 · Data augmentation plays a critical role in improving model performance across various domains, but it becomes challenging with graph data due to their complex and irregular structure. To address this issue, we propose EPIC (Edit Path Interpolation via learnable Cost), a novel interpolation-based method for augmenting graph datasets. To interpolate between two graphs lying in an irregular domain, EPIC leverages the concept of graph edit distance, constructing an edit path that represents the transformation process between two graphs via edit operations. Moreover, our method introduces a context-sensitive cost model that accounts for the importance of specific edit operations formulated through a learning framework. This allows for a more nuanced transformation process, where the edit distance is not merely count-based but reflects meaningful graph attributes. With randomly sampled graphs from the edit path, we enrich the training set to enhance the generalization capability of classification models. Experimental evaluations across several benchmark datasets demonstrate that our approach outperforms existing augmentation techniques in many tasks. EPIC IJCAI 2024 Enhancing Sample Efficiency in Black-box Combinatorial Optimization via Symmetric Replay Training · Hyeonah Kim, Minsu Kim, Sungsoo Ahn, Jinkyoo Park · ICML 2024 · Deep reinforcement learning (DRL) has significantly advanced the field of combinatorial optimization (CO). However, its practicality is hindered by the necessity for a large number of reward evaluations, especially in scenarios involving computationally intensive function assessments. To enhance the sample efficiency, we propose a simple but effective method, called symmetric replay training (SRT), which can be easily integrated into various DRL methods. Our method leverages high-reward samples to encourage exploration of the under-explored symmetric regions without additional online interactions - free. Through replay training, the policy is trained to maximize the likelihood of the symmetric trajectories of discovered high-rewarded samples. Experimental results demonstrate the consistent improvement of our method in sample efficiency across diverse DRL methods applied to real-world tasks, such as molecular optimization and hardware design. SRT ICML 2024 Graph Generation with K^2 Trees · Yunhui Jang, Dongwoo Kim, Sungsoo Ahn · ICLR 2024 · Generating graphs from a target distribution is a significant challenge across many domains, including drug discovery and social network analysis. In this work, we introduce a novel graph generation method leveraging K²-tree representation, originally designed for lossless graph compression. The K²-tree representation encompasses inherent hierarchy while enabling compact graph generation. In addition, we make contributions by (1) presenting a sequential K²-treerepresentation that incorporates pruning, flattening, and tokenization processes and (2) introducing a Transformer-based architecture designed to generate the sequence by incorporating a specialized tree positional encoding scheme. Finally, we extensively evaluate our algorithm on four general and two molecular graph datasets to confirm its superiority for graph generation. HGGT ICLR 2024 A Closer Look at the Intervention Procedure of Concept Bottleneck Models · Sungbin Shin, Yohan Jo, Sungsoo Ahn, Namhoon Lee · ICML 2023 · Concept bottleneck models (CBMs) are a class of interpretable neural network models that predict the target response of a given input based on its high-level concepts. Unlike the standard end-to-end models, CBMs enable domain experts to intervene on the predicted concepts and rectify any mistakes at test time, so that more accurate task predictions can be made at the end. While such intervenability provides a powerful avenue of control, many aspects of the intervention procedure remain rather unexplored. In this work, we develop various ways of selecting intervening concepts to improve the intervention effectiveness and conduct an array of in-depth analyses as to how they evolve under different circumstances. Specifically, we find that an informed intervention strategy can reduce the task error more than ten times compared to the current baseline under the same amount of intervention counts in realistic settings, and yet, this can vary quite significantly when taking into account different intervention granularity. We verify our findings through comprehensive evaluations, not only on the standard real datasets, but also on synthetic datasets that we generate based on a set of different causal graphs. We further discover some major pitfalls of the current practices which, without a proper addressing, raise concerns on reliability and fairness of the intervention procedure. Concept interventions ICML 2023 Diffusion Probabilistic Models for Structured Node Classification · Hyosoon Jang, Seonghyun Park, Sangwoo Mo, Sungsoo Ahn · NeurIPS 2023 · This paper studies structured node classification on graphs, where the predictions should consider dependencies between the node labels. In particular, we focus on solving the problem for partially labeled graphs where it is essential to incorporate the information in the known label for predicting the unknown labels. To address this issue, we propose a novel framework leveraging the diffusion probabilistic model for structured node classification (DPM-SNC). At the heart of our framework is the extraordinary capability of DPM-SNC to (a) learn a joint distribution over the labels with an expressive reverse diffusion process and (b) make predictions conditioned on the known labels utilizing manifold-constrained sampling. Since the DPMs lack training algorithms for partially labeled data, we design a novel training algorithm to apply DPMs, maximizing a new variational lower bound. We also theoretically analyze how DPMs benefit node classification by enhancing the expressive power of GNNs based on proposing AGG-WL, which is strictly more powerful than the classic 1-WL test. We extensively verify the superiority of our DPM-SNC in diverse scenarios, which include not only the transductive setting on partially labeled graphs but also the inductive setting and unlabeled graphs. DPM-SNC NeurIPS 2023 Imitating Graph-Based Planning with Goal-Conditioned Policies · Junsu Kim, Younggyo Seo, Sungsoo Ahn, Kyunghwan Son, Jinwoo Shin · ICLR 2023 · Recently, graph-based planning algorithms have gained much attention to solve goal-conditioned reinforcement learning (RL) tasks: they provide a sequence of subgoals to reach the target-goal, and the agents learn to execute subgoal-conditioned policies. However, the sample-efficiency of such RL schemes still remains a challenge, particularly for long-horizon tasks. To address this issue, we present a simple yet effective self-imitation scheme which distills a subgoal-conditioned policy into the target-goal-conditioned policy. Our intuition here is that to reach a target-goal, an agent should pass through a subgoal, so target-goal- and subgoal- conditioned policies should be similar to each other. We also propose a novel scheme of stochastically skipping executed subgoals in a planned path, which further improves performance. Unlike prior methods that only utilize graph-based planning in an execution phase, our method transfers knowledge from a planner along with a graph into policy learning. We empirically show that our method can significantly boost the sample-efficiency of the existing goal-conditioned RL methods under various long-horizon control tasks. PIG ICLR 2023 Learning Debiased Classifier with Biased Committee · Nayeong Kim, Sehyun Hwang, Sungsoo Ahn, Jaesik Park, Suha Kwak · NeurIPS 2022 · Neural networks are prone to be biased towards spurious correlations between classes and latent attributes exhibited in a major portion of training data, which ruins their generalization capability. We propose a new method for training debiased classifiers with no spurious attribute label. The key idea is to employ a committee of classifiers as an auxiliary module that identifies bias-conflicting data, i.e., data without spurious correlation, and assigns large weights to them when training the main classifier. The committee is learned as a bootstrapped ensemble so that a majority of its classifiers are biased as well as being diverse, and intentionally fail to predict classes of bias-conflicting data accordingly. The consensus within the committee on prediction difficulty thus provides a reliable cue for identifying and weighting bias-conflicting data. Moreover, the committee is also trained with knowledge transferred from the main classifier so that it gradually becomes debiased along with the main classifier and emphasizes more difficult data as training progresses. On five real-world datasets, our method outperforms prior arts using no spurious attribute label like ours and even surpasses those relying on bias labels occasionally. LWBC NeurIPS 2022 Disentangling Sources of Risk for Distributional Multi-Agent Reinforcement Learning · Kyunghwan Son, Junsu Kim, Sungsoo Ahn, Roben Delos Reyes, Yung Yi, Jinwoo Shin · ICML 2022 · In cooperative multi-agent reinforcement learning, the outcomes of agent-wise policies are highly stochastic due to the two sources of risk: (a) random actions taken by teammates and (b) random transition and rewards. Although the two sources have very distinct characteristics, existing frameworks are insufficient to control the risk-sensitivity of agent-wise policies in a disentangled manner. To this end, we propose Disentangled RIsk-sensitive Multi-Agent reinforcement learning (DRIMA) to separately access the risk sources. For example, our framework allows an agent to be optimistic with respect to teammates (who can prosocially adapt) but more risk-neutral with respect to the environment (which does not adapt). Our experiments demonstrate that DRIMA significantly outperforms prior state-of-the-art methods across various scenarios in the StarCraft Multi-agent Challenge environment. Notably, DRIMA shows robust performance where prior methods learn only a highly suboptimal policy, regardless of reward shaping, exploration scheduling, and noisy (random or adversarial) agents. DRIMA ICML 2022 What Makes Better Augmentation Strategies? Augment Difficult but Not Too Different · Jaehyung Kim, Dongyeop Kang, Sungsoo Ahn, Jinwoo Shin · ICLR 2022 · The practice of data augmentation has been extensively used to boost the performance of deep neural networks for various NLP tasks. It is more effective when only a limited number of labeled samples is available, e.g., low-data or class-imbalanced regimes. Most current augmentation techniques rely on parameter tuning or inherent randomness; hence, their effectiveness largely varies on the tasks. To efficiently find the best augmentation strategy for each task, learning data augmentation policy is a promising solution, but the question of what makes a good augmentation in NLP tasks and how to design the reward function for learning a good policy remains under-explored. To answer this, we hypothesize that good data augmentation should construct more diverse and challenging samples for providing informative training signals, while avoiding the risk of losing the semantics of original samples. Therefore, we design a novel reward function for updating the augmentation policy to construct difficult but not too different samples (DND). Particularly, we jointly optimize a data augmentation policy while training the model, to construct the augmented samples with low confidence but a high semantic similarity with original ones. In addition, we introduce a sample re-weighting scheme to focus on difficult augmented samples after the original ones are learned confidently for more effective learning from the augmented ones. Our learning-based augmentation outperforms the recent state-of-the-art augmentation schemes on various text classification tasks and GLUE benchmark by successfully discovering the effective augmentations for each task. Remarkably, our method is more effective on the challenging low-data and class-imbalanced regimes, and the learned augmentation policy is well-transferable to the different tasks and models. DND ICLR 2022 Spanning Tree-based Graph Generation for Molecules · Sungsoo Ahn, Binghong Chen, Tianzhe Wang, Le Song · ICLR 2022 · In this paper, we explore the problem of generating molecules using deep neural networks, which has recently gained much interest in chemistry. To this end, we propose a spanning tree-based graph generation (STGG) framework based on formulating molecular graph generation as a construction of a spanning tree and the residual edges. Such a formulation exploits the sparsity of molecular graphs and allows using compact tree-constructive operations to define the molecular graph connectivity. Based on the intermediate graph structure of the construction process, our framework can constrain its generation to molecular graphs that satisfy the chemical valence rules. We also newly design a Transformer architecture with tree-based relative positional encodings for realizing the tree construction procedure. Experiments on QM9, ZINC250k, and MOSES benchmarks verify the effectiveness of the proposed framework in metrics such as validity, Frechet ChemNet distance, and fragment similarity. We also demonstrate the usefulness of STGG in maximizing penalized LogP value of molecules. STGG ICLR 2022 RoMA: Robust Model Adaptation for Offline Model-based Optimization · Sihyun Yu, Sungsoo Ahn, Le Song, Jinwoo Shin · NeurIPS 2021 · We consider the problem of searching an input maximizing a black-box objective function given a static dataset of input-output queries. A popular approach to solving this problem is maintaining a proxy model, e.g., a deep neural network (DNN), that approximates the true objective function. Here, the main challenge is how to avoid adversarially optimized inputs during the search, i.e., the inputs where the DNN highly overestimates the true objective function. To handle the issue, we propose a new framework, coined robust model adaptation (RoMA), based on gradient-based optimization of inputs over the DNN. Specifically, it consists of two steps: (a) a pre-training strategy to robustly train the proxy model and (b) a novel adaptation procedure of the proxy model to have robust estimates for a specific set of candidate solutions. At a high level, our scheme utilizes the local smoothness prior to overcome the brittleness of the DNN. Experiments under various tasks show the effectiveness of RoMA compared with previous methods, obtaining state-of-the-art results, e.g., RoMA outperforms all at 4 out of 6 tasks and achieves runner-up results at the remaining tasks. RoMA NeurIPS 2021 Self-Improved Retrosynthetic Planning · Junsu Kim, Sungsoo Ahn, Hankook Lee, Jinwoo Shin · ICML 2021 · Retrosynthetic planning is a fundamental problem in chemistry for finding a pathway of reactions to synthesize a target molecule. Recently, search algorithms have shown promising results for solving this problem by using deep neural networks (DNNs) to expand their candidate solutions, i.e., adding new reactions to reaction pathways. However, the existing works on this line are suboptimal; the retrosynthetic planning problem requires the reaction pathways to be (a) represented by real-world reactions and (b) executable using "building block" molecules, yet the DNNs expand reaction pathways without fully incorporating such requirements. Motivated by this, we propose an end-to-end framework for directly training the DNNs towards generating reaction pathways with the desirable properties. Our main idea is based on a self-improving procedure that trains the model to imitate successful trajectories found by itself. We also propose a novel reaction augmentation scheme based on a forward reaction model. Our experiments demonstrate that our scheme significantly improves the success rate of solving the retrosynthetic problem from 86.84% to 96.32% while maintaining the performance of DNN for predicting valid reactions. Self-improved retro ICML 2021 RetCL: A Selection-based Approach for Retrosynthesis via Contrastive Learning · Hankook Lee, Sungsoo Ahn, Seung-Woo Seo, You Young Song, Eunho Yang, Sung-Ju Hwang, Jinwoo Shin · IJCAI 2021 · Retrosynthesis, of which the goal is to find a set of reactants for synthesizing a target product, is an emerging research area of deep learning. While the existing approaches have shown promising results, they currently lack the ability to consider availability (e.g., stability or purchasability) of the reactants or generalize to unseen reaction templates (i.e., chemical reaction rules). In this paper, we propose a new approach that mitigates the issues by reformulating retrosynthesis into a selection problem of reactants from a candidate set of commercially available molecules. To this end, we design an efficient reactant selection framework, named RetCL (retrosynthesis via contrastive learning), for enumerating all of the candidate molecules based on selection scores computed by graph neural networks. For learning the score functions, we also propose a novel contrastive training scheme with hard negative mining. Extensive experiments demonstrate the benefits of the proposed selection-based approach. For example, when all 671k reactants in the USPTO database are given as candidates, our RetCL achieves top-1 exact match accuracy of 71.3% for the USPTO-50k benchmark, while a recent transformer-based approach achieves 59.6%. We also demonstrate that RetCL generalizes well to unseen templates in various settings in contrast to template-based approaches. RetCL IJCAI 2021 Layer-adaptive sparsity for the Magnitude-based Pruning · Jaeho Lee, Sejun Park, Sangwoo Mo, Sungsoo Ahn, Jinwoo Shin · ICLR 2021 · Recent discoveries on neural network pruning reveal that, with a carefully chosen layerwise sparsity, a simple magnitude-based pruning achieves state-of-the-art tradeoff between sparsity and performance. However, without a clear consensus on "how to choose," the layerwise sparsities are mostly selected algorithm-by-algorithm, often resorting to handcrafted heuristics or an extensive hyperparameter search. To fill this gap, we propose a novel importance score for global pruning, coined layer-adaptive magnitude-based pruning (LAMP) score; the score is a rescaled version of weight magnitude that incorporates the model-level ℓ₂ distortion incurred by pruning, and does not require any hyperparameter tuning or heavy computation. Under various image classification setups, LAMP consistently outperforms popular existing schemes for layerwise sparsity selection. Furthermore, we observe that LAMP continues to outperform baselines even in weight-rewinding setups, while the connectivity-oriented layerwise sparsity (the strongest baseline overall) performs worse than a simple global magnitude-based pruning in this case. Code: https://github.com/jaeho-lee/layer-adaptive-sparsity LAMP ICLR 2021 Learning from Failure: Training Debiased Classifier from Biased Classifier · Junhyun Nam, Hyuntak Cha, Sungsoo Ahn, Jaeho Lee, Jinwoo Shin · NeurIPS 2020 · Neural networks often learn to make predictions that overly rely on spurious correlation existing in the dataset, which causes the model to be biased. While previous work tackles this issue by using explicit labeling on the spuriously correlated attributes or presuming a particular bias type, we instead utilize a cheaper, yet generic form of human knowledge, which can be widely applicable to various types of bias. We first observe that neural networks learn to rely on the spurious correlation only when it is "easier" to learn than the desired knowledge, and such reliance is most prominent during the early phase of training. Based on the observations, we propose a failure-based debiasing scheme by training a pair of neural networks simultaneously. Our main idea is twofold; (a) we intentionally train the first network to be biased by repeatedly amplifying its "prejudice", and (b) we debias the training of the second network by focusing on samples that go against the prejudice of the biased network in (a). Extensive experiments demonstrate that our method significantly improves the training of the network against various types of biases in both synthetic and real-world datasets. Surprisingly, our framework even occasionally outperforms the debiasing methods requiring explicit supervision of the spuriously correlated attributes. LfF NeurIPS 2020 Guiding Deep Molecular Optimization with Genetic Exploration · Sungsoo Ahn, Junsu Kim, Hankook Lee, Jinwoo Shin · NeurIPS 2020 · De novo molecular design attempts to search over the chemical space for molecules with the desired property. Recently, deep learning has gained considerable attention as a promising approach to solve the problem. In this paper, we propose genetic expert-guided learning (GEGL), a simple yet novel framework for training a deep neural network (DNN) to generate highly-rewarding molecules. Our main idea is to design a "genetic expert improvement" procedure, which generates high-quality targets for imitation learning of the DNN. Extensive experiments show that GEGL significantly improves over state-of-the-art methods. For example, GEGL manages to solve the penalized octanol-water partition coefficient optimization with a score of 31.40, while the best-known score in the literature is 27.22. Besides, for the GuacaMol benchmark with 20 tasks, our method achieves the highest score for 19 tasks, in comparison with state-of-the-art methods, and newly obtains the perfect score for three tasks. GEGL NeurIPS 2020 Learning What to Defer for Maximum Independent Sets · Sungsoo Ahn, Younggyo Seo, Jinwoo Shin · ICML 2020 · Designing efficient algorithms for combinatorial optimization appears ubiquitously in various scientific fields. Recently, deep reinforcement learning (DRL) frameworks have gained considerable attention as a new approach: they can automate the design of a solver while relying less on sophisticated domain knowledge of the target problem. However, the existing DRL solvers determine the solution using a number of stages proportional to the number of elements in the solution, which severely limits their applicability to large-scale graphs. In this paper, we seek to resolve this issue by proposing a novel DRL scheme, coined learning what to defer (LwD), where the agent adaptively shrinks or stretch the number of stages by learning to distribute the element-wise decisions of the solution at each stage. We apply the proposed framework to the maximum independent set (MIS) problem, and demonstrate its significant improvement over the current state-of-the-art DRL scheme. We also show that LwD can outperform the conventional MIS solvers on large-scale graphs having millions of vertices, under a limited time budget. LwD ICML 2020 Variational Information Distillation for Knowledge Transfer · Sungsoo Ahn, Shell Xu Hu, Andreas Damianou, Neil D. Lawrence, Zhenwen Dai · CVPR 2019 · Transferring knowledge from a teacher neural network pretrained on the same or a similar task to a student neural network can significantly improve the performance of the student neural network. Existing knowledge transfer approaches match the activations or the corresponding hand-crafted features of the teacher and the student networks. We propose an information-theoretic framework for knowledge transfer which formulates knowledge transfer as maximizing the mutual information between the teacher and the student networks. We compare our method with existing knowledge transfer methods on both knowledge distillation and transfer learning tasks and show that our method consistently outperforms existing methods. We further demonstrate the strength of our method on knowledge transfer across heterogeneous network architectures by transferring knowledge from a convolutional neural network (CNN) to a multi-layer perceptron (MLP) on CIFAR-10. The resulting MLP significantly outperforms the-state-of-the-art methods and it achieves similar performance to the CNN with a single convolutional layer. VID CVPR 2019 Bucket-Renormalization for Approximate Inference · Sungsoo Ahn, Michael Chertkov, Adrian Weller, Jinwoo Shin · ICML 2018 · JSTAT 2019 · Probabilistic graphical models are a key tool in machine learning applications. Computing the partition function, i.e., normalizing constant, is a fundamental task of statistical inference but it is generally computationally intractable, leading to extensive study of approximation methods. Iterative variational methods are a popular and successful family of approaches. However, even state of the art variational methods can return poor results or fail to converge on difficult instances. In this paper, we instead consider computing the partition function via sequential summation over variables. We develop robust approximate algorithms by combining ideas from mini-bucket elimination with tensor network and renormalization group methods from statistical physics. The resulting "convergence-free" methods show good empirical performance on both synthetic and real-world benchmark models, even for difficult instances. MBR / GBR ICML 2018 Maximum Weight Matching using Odd-sized Cycles: Max-Product Belief Propagation and Half-Integrality · Sungsoo Ahn, Michael Chertkov, Andrew E. Gelfand, Sejun Park, Jinwoo Shin · IEEE TIT 2018 · We study the Maximum Weight Matching (MWM) problem for general graphs through the max-product Belief Propagation (BP) and related Linear Programming (LP). The BP approach provides distributed heuristics for finding the Maximum A Posteriori (MAP) assignment in a joint probability distribution represented by a Graphical Model (GM) and respective LPs can be considered as continuous relaxations of the discrete MAP problem. It was recently shown that a BP algorithm converges to the correct MWM assignment under a simple GM formulation of MAP/MWM as long as the corresponding LP relaxation is tight. First, under the motivation for forcing the tightness condition, we consider a new GM formulation of MWM, say C-GM, using non-intersecting odd-sized cycles in the graph: the new corresponding LP relaxation, say C-LP, becomes tight for more MWM instances. However, the tightness of C-LP now does not guarantee such convergence and correctness of the new BP on C-GM. To address the issue, we introduce a novel graph transformation applied to C-GM, which results in another GM formulation of MWM, and prove that the respective BP on it converges to the correct MAP/MWM assignment as long as C-LP is tight. Finally, we also show that C-LP always has half-integral solutions, which leads to an efficient BP-based MWM heuristic consisting of making sequential, `cutting plane', modifications to the underlying GM. Our experiments show that this BP-based cutting plane heuristic performs as well as that based on traditional LP solvers. Odd-cycle BP IEEE TIT 2018 Gauged Mini-Bucket Elimination for Approximate Inference · Sungsoo Ahn, Michael Chertkov, Jinwoo Shin, Adrian Weller · AISTATS 2018 · Computing the partition function Z of a discrete graphical model is a fundamental inference challenge. Since this is computationally intractable, variational approximations are often used in practice. Recently, so-called gauge transformations were used to improve variational lower bounds on Z. In this paper, we propose a new gauge-variational approach, termed WMBE-G, which combines gauge transformations with the weighted mini-bucket elimination (WMBE) method. WMBE-G can provide both upper and lower bounds on Z, and is easier to optimize than the prior gauge-variational algorithm. We show that WMBE-G strictly improves the earlier WMBE approximation for symmetric models including Ising models with no magnetic field. Our experimental results demonstrate the effectiveness of WMBE-G even for generic, nonsymmetric models. WMBE-G AISTATS 2018 Gauging Variational Inference · Sungsoo Ahn, Michael Chertkov, Jinwoo Shin · NeurIPS 2017 · JSTAT 2019 · Computing partition function is the most important statistical inference task arising in applications of Graphical Models (GM). Since it is computationally intractable, approximate methods have been used to resolve the issue in practice, where mean-field (MF) and belief propagation (BP) are arguably the most popular and successful approaches of a variational type. In this paper, we propose two new variational schemes, coined Gauged-MF (G-MF) and Gauged-BP (G-BP), improving MF and BP, respectively. Both provide lower bounds for the partition function by utilizing the so-called gauge transformation which modifies factors of GM while keeping the partition function invariant. Moreover, we prove that both G-MF and G-BP are exact for GMs with a single loop of a special structure, even though the bare MF and BP perform badly in this case. Our extensive experiments, on complete GMs of relatively small size and on large GM (up-to 300 variables) confirm that the newly proposed algorithms outperform and generalize MF and BP. G-MF / G-BP NeurIPS 2017 Synthesis of MCMC and Belief Propagation · Sungsoo Ahn, Michael Chertkov, Jinwoo Shin · NeurIPS 2016 · Markov Chain Monte Carlo (MCMC) and Belief Propagation (BP) are the most popular algorithms for computational inference in Graphical Models (GM). In principle, MCMC is an exact probabilistic method which, however, often suffers from exponentially slow mixing. In contrast, BP is a deterministic method, which is typically fast, empirically very successful, however in general lacking control of accuracy over loopy graphs. In this paper, we introduce MCMC algorithms correcting the approximation error of BP, i.e., we provide a way to compensate for BP errors via a consecutive BP-aware MCMC. Our framework is based on the Loop Calculus (LC) approach which allows expressing the BP error as a sum of weighted generalized loops. Although the full series is computationally intractable, it is known that a truncated series, summing up all 2-regular loops, is computable in polynomial-time for planar pair-wise binary GMs and it also provides a highly accurate approximation empirically. Motivated by this, we first propose a polynomial-time approximation MCMC scheme for the truncated series of general (non-planar) pair-wise binary models. Our main idea here is to use the Worm algorithm, known to provide fast mixing in other (related) problems, and then design an appropriate rejection scheme to sample 2-regular loops. Furthermore, we also design an efficient rejection-free MCMC scheme for approximating the full series. The main novelty underlying our design is in utilizing the concept of cycle basis, which provides an efficient decomposition of the generalized loops. In essence, the proposed MCMC schemes run on transformed GM built upon the non-trivial BP solution, and our experiments show that this synthesis of BP and MCMC outperforms both direct MCMC and bare BP schemes. MCMC + BP NeurIPS 2016 Minimum Weight Perfect Matching via Blossom Belief Propagation · Sungsoo Ahn, Sejun Park, Michael Chertkov, Jinwoo Shin · NeurIPS 2015 · Max-product Belief Propagation (BP) is a popular message-passing algorithm for computing a Maximum-A-Posteriori (MAP) assignment over a distribution represented by a Graphical Model (GM). It has been shown that BP can solve a number of combinatorial optimization problems including minimum weight matching, shortest path, network flow and vertex cover under the following common assumption: the respective Linear Programming (LP) relaxation is tight, i.e., no integrality gap is present. However, when LP shows an integrality gap, no model has been known which can be solved systematically via sequential applications of BP. In this paper, we develop the first such algorithm, coined Blossom-BP, for solving the minimum weight matching problem over arbitrary graphs. Each step of the sequential algorithm requires applying BP over a modified graph constructed by contractions and expansions of blossoms, i.e., odd sets of vertices. Our scheme guarantees termination in O(n^2) of BP runs, where n is the number of vertices in the original graph. In essence, the Blossom-BP offers a distributed version of the celebrated Edmonds' Blossom algorithm by jumping at once over many sub-steps with a single BP. Moreover, our result provides an interpretation of the Edmonds' algorithm as a sequence of LPs. Blossom-BP NeurIPS 2015
scTrilemma
Cells
NeurIPS 2026
Balances biological identity, nuisance invariance, and expression fidelity in single-cell embeddings.
Packora
MaterialsMolecules
arXiv 2026
Generates molecular crystal structures while separating candidate generation from downstream ranking.
MaskGXT / HACO
Materials
arXiv 2026
An AI co-scientist develops masked generative models for crystal structure prediction.
MADField
Materials
arXiv 2026
Estimates adsorption equilibrium as a spatial density field instead of repeatedly simulating individual gas particles.
TriProRep
Proteins / DNA
NeurIPS 2026
Learns protein representations that support structure prediction rather than only function annotation.
VibeProteinBench
Proteins / DNA
arXiv 2026
Evaluates protein recognition, engineering, and generation through natural-language requests.
MCFlow
Materials
ICML 2026
Unifies crystal generation tasks that previously required task-specific models.
AdaPert
Cells
ICML 2026
Predicts perturbation-specific transcriptional responses without collapsing to an average cell response.
QHFlow2
Electronic structure
ICML 2026
Tests predicted electronic Hamiltonians by the energies and forces they actually produce.
RMF
General MLProteins / DNA
ICML 2026
Learns generative flow maps directly on manifolds for rapid biological design.
PBio-Agent
Cells
arXiv 2026
Predicts gene-regulation directions under chemical and genetic perturbations with structured biological reasoning.
AtomMOF
MaterialsMolecules
NeurIPS 2026
Predicts MOF and adsorbate structures directly from the graphs of their building blocks.
CatFlow
MaterialsMolecules
ICML 2026
Co-generates catalyst surfaces and adsorbate configurations as a coupled structure.
INDIBATOR
Molecules
arXiv 2026
Grounds molecular-discovery debates in individualized scientific and molecular histories.
Co-folding reps
MoleculesProteins / DNA
NeurIPS 2026
Tests whether protein–ligand co-folding can supply useful representations for standalone small molecules.
DNAChunker
Proteins / DNA
ICML 2026
Learns the segmentation of DNA jointly with a masked language model.
BioEmu-CV
Proteins / DNA
ICLR 2026
Learns slow collective variables from time-lagged conditioning of a pretrained protein ensemble generator.
MT-Mol
Molecules
EMNLP 2025
Optimizes molecules with specialized agents that use chemistry tools to inspect proposals.
SGDS
MoleculesGeneral ML
NeurIPS 2025
Improves neural sampling when energy calls are expensive and landscapes contain difficult-to-reach modes.
EGM
General ML
NeurIPS 2025
Trains neural samplers directly from energy functions on continuous, discrete, or mixed spaces.
QHFlow
Electronic structure
NeurIPS 2025
Predicts structured electronic Hamiltonians to reduce expensive self-consistent-field computations.
CORE-PO
General ML
EMNLP 2025
Improves reasoning self-training by evaluating reasoning paths rather than only final answers.
MOFFlow-2
MaterialsMolecules
NeurIPS 2025
Generates MOF building blocks and assembles flexible structures with novel chemistry.
MELD
Molecules
ICLR 2026
Addresses collisions between masked-diffusion trajectories of distinct molecular graphs.
CleanMol
Molecules
EMNLP 2025
Teaches language models to recover molecular graph information from SMILES strings.
VS / AVI
General ML
ICLR 2026
Infers which individual steps in a reasoning chain are incorrect.
CIP
General ML
ACL 2025
Uses causal influence diagrams to guide safer language-model agent decisions.
MSR
Molecules
ACL 2025
Improves language-model molecular understanding through explicit structural sketches.
MOFFlow
Materials
ICLR 2025
Predicts MOF structures by assembling metal nodes and organic linkers as rigid building blocks.
RxnFlow
Molecules
ICLR 2025
Designs diverse molecules through explicit synthetic reaction pathways.
ReBind
Molecules
ICLR 2025
Improves conformation prediction for atoms whose geometry depends strongly on non-bonded interactions.
Div-SFT+RL
Molecules
arXiv 2024
Aligns a language model with structural diversity across generated molecules.
Adaptive Teachers
MoleculesGeneral MLProteins / DNA
ICLR 2025
Directs amortized samplers toward states their current training policy misses.
iEFM
General ML
arXiv 2024
Learns a continuous generator from an energy function when target samples are unavailable.
RL4CO
General ML
KDD 2025
Provides a common framework for comparing reinforcement-learning solvers for combinatorial problems.
HoliMol
Molecules
TMLR 2024
Connects global molecular graph features to chemically meaningful local fragments.
TPS-DPS
MoleculesProteins / DNA
ICLR 2025
Samples molecular transition paths without requiring manually designed collective variables.
ASSD
Proteins / DNA
TMLR 2024
Separates antibody sequence generation from structure prediction to improve architecture flexibility.
PBP-GFN
MoleculesGeneral MLProteins / DNA
NeurIPS 2024
Addresses under-exploitation of high-reward objects in GFlowNet training.
UCom2
General ML
ICML 2024
Constructs differentiable objectives and rounding rules for constrained discrete optimization.
Multi-bias learning
General ML
ICML 2024
Handles multiple spurious correlations whose mitigation objectives can conflict.
BEAG
General ML
ICML 2024
Explores long-horizon goal spaces while recording both achieved and unattained subgoals.
HNeR-S
Weather / cosmos
ICML 2024
Represents weather and cosmic microwave background signals on the sphere.
GPWNO
Electronic structure
ICML 2024
Predicts electronic density using complementary local and global basis functions.
GEEL
Molecules
ICLR 2024
Makes sparse graph generation scale with edges rather than full adjacency matrices.
NBA-GNN
General MLProteins / DNA
TMLR 2024
Reduces redundant message flows that revisit the immediately preceding graph node.
LED-GFN
MoleculesGeneral MLProteins / DNA
ICLR 2024
Provides local training signals when intermediate object energies are costly or misleading.
LS-GFN
MoleculesGeneral MLProteins / DNA
ICLR 2024
Combines global sampling with refinement of promising local neighborhoods.
Wave-GD
MoleculesGeneral ML
NeurIPS 2023
Preserves graph-frequency structure during joint node-and-edge generation.
BootGen
Proteins / DNA
NeurIPS 2023
Designs biological sequences from a fixed offline dataset of scored examples.
EPIC
MoleculesGeneral MLProteins / DNA
IJCAI 2024
Interpolates between graphs through meaningful edit operations for data augmentation.
SRT
Molecules
ICML 2024
Makes black-box combinatorial optimization more sample-efficient when reward calls are costly.
HGGT
Molecules
ICLR 2024
Generates graphs through a compact hierarchical representation of their adjacency structure.
Concept interventions
General ML
ICML 2023
Examines how concept bottleneck interventions should be selected and applied.
DPM-SNC
General MLProteins / DNA
NeurIPS 2023
Predicts graph-node labels jointly rather than independently.
PIG
General ML
ICLR 2023
Transfers graph-planner knowledge into a goal-conditioned control policy.
LWBC
General ML
NeurIPS 2022
Finds bias-conflicting training examples without spurious-attribute labels.
DRIMA
General ML
ICML 2022
Separates uncertainty from teammates' actions and uncertainty from environmental dynamics.
DND
General ML
ICLR 2022
Learns text augmentations that are challenging while retaining original semantics.
STGG
Molecules
ICLR 2022
Generates molecular graphs by constructing a spanning tree and its remaining edges.
RoMA
General MLProteins / DNA
NeurIPS 2021
Optimizes a learned objective while limiting exploitation of out-of-distribution model errors.
Self-improved retro
Molecules
ICML 2021
Trains retrosynthetic reaction proposals to support complete executable search routes.
RetCL
Molecules
IJCAI 2021
Selects available reactants rather than generating unrestricted reactant descriptions.
LAMP
General ML
ICLR 2021
Allocates sparsity across neural-network layers without extensive layerwise tuning.
LfF
General ML
NeurIPS 2020
Debiases a classifier by emphasizing examples that an intentionally biased model gets wrong.
GEGL
Molecules
NeurIPS 2020
Uses domain-specific genetic search to improve a neural molecular generator.
LwD
General ML
ICML 2020
Solves large maximum-independent-set problems with an adaptive number of decision stages.
VID
General ML
CVPR 2019
Transfers knowledge between neural networks by maximizing shared information.
MBR / GBR
Graphical models
ICML 2018
Approximates graphical-model partition functions through sequential variable elimination.
Odd-cycle BP
Graphical models
IEEE TIT 2018
Reformulates maximum-weight matching so belief propagation can exploit odd-cycle constraints.
WMBE-G
Graphical models
AISTATS 2018
Combines gauge transformations with bounded-complexity variable elimination.
G-MF / G-BP
Graphical models
NeurIPS 2017
Improves partition-function approximations by transforming graphical-model factors.
MCMC + BP
Graphical models
NeurIPS 2016
Corrects belief-propagation partition estimates using Monte Carlo on a transformed loop model.
Blossom-BP
Graphical models
NeurIPS 2015
Solves minimum-weight perfect matching through successive belief-propagation problems.
Domains Graphical models General ML Molecules Proteins / DNA Electronic structure Weather / cosmos Materials Cells

Figure inspired by Kirill Neklyudov.