Jongmin Lee

Jongmin Lee, Cosmin Paduraru, Daniel J. Mankowitz et al. · deepmind

We consider the offline constrained reinforcement learning (RL) problem, in which the agent aims to compute a policy that maximizes expected return while satisfying given cost constraints, learning only from a pre-collected dataset. This problem setting is appealing in many real-world scenarios, where direct interaction with the environment is costly or risky, and where the resulting policy should comply with safety constraints. However, it is challenging to compute a policy that guarantees satisfying the cost constraints in the offline RL setting, since the off-policy evaluation inherently has an estimation error. In this paper, we present an offline constrained RL algorithm that optimizes the policy in the space of the stationary distribution. Our algorithm, COptiDICE, directly estimates the stationary distribution corrections of the optimal policy with respect to returns, while constraining the cost upper bound, with the goal of yielding a cost-conservative policy for actual constraint satisfaction. Experimental results show that COptiDICE attains better policies in terms of constraint satisfaction and return-maximization, outperforming baseline algorithms.

Carmelo Sferrazza, Dun-Ming Huang, Fangchen Liu et al.

In recent years, the transformer architecture has become the de facto standard for machine learning algorithms applied to natural language processing and computer vision. Despite notable evidence of successful deployment of this architecture in the context of robot learning, we claim that vanilla transformers do not fully exploit the structure of the robot learning problem. Therefore, we propose Body Transformer (BoT), an architecture that leverages the robot embodiment by providing an inductive bias that guides the learning process. We represent the robot body as a graph of sensors and actuators, and rely on masked attention to pool information throughout the architecture. The resulting architecture outperforms the vanilla transformer, as well as the classical multilayer perceptron, in terms of task completion, scaling properties, and computational efficiency when representing either imitation or reinforcement learning policies. Additional material including the open-source code is available at https://sferrazza.cc/bot_site.

Minkyung Kwon, Jinhyeok Choi, Youngjin Shin et al.

We present MORPHOS, a novel autoregressive framework that generates dynamic 3D assets from videos across diverse representations, including meshes, 3D Gaussians, and radiance fields. Existing methods are typically limited to a single representation, struggle to model topological changes, or fail to maintain temporal consistency over long videos. To address these limitations, we introduce the Temporal Structured Latents (T-SLAT), a unified 4D representation that jointly encodes geometry and appearance along the temporal dimension. Leveraging T-SLAT, MORPHOS autoregressively generates dynamic 3D assets via causal attention, conditioning each frame on its preceding history to ensure temporal consistency while handling evolving topologies. We also propose a temporal-structural augmentation to mitigate error accumulation in autoregressive generation. MORPHOS achieves state-of-the-art performance in appearance and competitive results in geometry across multiple benchmarks, demonstrating superior generalization across various representations and robustness in long-horizon generation.

Jongmin Lee, Byungjin Kim, Minsu Cho

Detecting robust keypoints from an image is an integral part of many computer vision problems, and the characteristic orientation and scale of keypoints play an important role for keypoint description and matching. Existing learning-based methods for keypoint detection rely on standard translation-equivariant CNNs but often fail to detect reliable keypoints against geometric variations. To learn to detect robust oriented keypoints, we introduce a self-supervised learning framework using rotation-equivariant CNNs. We propose a dense orientation alignment loss by an image pair generated by synthetic transformations for training a histogram-based orientation map. Our method outperforms the previous methods on an image matching benchmark and a camera pose estimation benchmark.

Jongmin Lee, Byungjin Kim, Seungwook Kim et al.

Extracting discriminative local features that are invariant to imaging variations is an integral part of establishing correspondences between images. In this work, we introduce a self-supervised learning framework to extract discriminative rotation-invariant descriptors using group-equivariant CNNs. Thanks to employing group-equivariant CNNs, our method effectively learns to obtain rotation-equivariant features and their orientations explicitly, without having to perform sophisticated data augmentations. The resultant features and their orientations are further processed by group aligning, a novel invariant mapping technique that shifts the group-equivariant features by their orientations along the group dimension. Our group aligning technique achieves rotation-invariance without any collapse of the group dimension and thus eschews loss of discriminability. The proposed method is trained end-to-end in a self-supervised manner, where we use an orientation alignment loss for the orientation estimation and a contrastive descriptor loss for robust local descriptors to geometric/photometric variations. Our method demonstrates state-of-the-art matching accuracy among existing rotation-invariant descriptors under varying rotation and also shows competitive results when transferred to the task of keypoint matching and camera pose estimation.

Jerome Revaud, Yohann Cabon, Romain Brégier et al.

Scene coordinates regression (SCR), i.e., predicting 3D coordinates for every pixel of a given image, has recently shown promising potential. However, existing methods remain limited to small scenes memorized during training, and thus hardly scale to realistic datasets and scenarios. In this paper, we propose a generalized SCR model trained once to be deployed in new test scenes, regardless of their scale, without any finetuning. Instead of encoding the scene coordinates into the network weights, our model takes as input a database image with some sparse 2D pixel to 3D coordinate annotations, extracted from e.g. off-the-shelf Structure-from-Motion or RGB-D data, and a query image for which are predicted a dense 3D coordinate map and its confidence, based on cross-attention. At test time, we rely on existing off-the-shelf image retrieval systems and fuse the predictions from a shortlist of relevant database images w.r.t. the query. Afterwards camera pose is obtained using standard Perspective-n-Point (PnP). Starting from selfsupervised CroCo pretrained weights, we train our model on diverse datasets to ensure generalizabilty across various scenarios, and significantly outperform other scene regression approaches, including scene-specific models, on multiple visual localization benchmarks. Finally, we show that the database representation of images and their 2D-3D annotations can be highly compressed with negligible loss of localization performance.

JongMin Lee, Yohann Cabon, Romain Brégier et al.

Existing learning-based methods for object pose estimation in RGB images are mostly model-specific or category based. They lack the capability to generalize to new object categories at test time, hence severely hindering their practicability and scalability. Notably, recent attempts have been made to solve this issue, but they still require accurate 3D data of the object surface at both train and test time. In this paper, we introduce a novel approach that can estimate in a single forward pass the pose of objects never seen during training, given minimum input. In contrast to existing state-of-the-art approaches, which rely on task-specific modules, our proposed model is entirely based on a transformer architecture, which can benefit from recently proposed 3D-geometry general pretraining. We conduct extensive experiments and report state-of-the-art one-shot performance on the challenging LINEMOD benchmark. Finally, extensive ablations allow us to determine good practices with this relatively new type of architecture in the field.

Hyunin Lee, Yuhao Ding, Jongmin Lee et al.

We first raise and tackle a ``time synchronization'' issue between the agent and the environment in non-stationary reinforcement learning (RL), a crucial factor hindering its real-world applications. In reality, environmental changes occur over wall-clock time ($t$) rather than episode progress ($k$), where wall-clock time signifies the actual elapsed time within the fixed duration $t \in [0, T]$. In existing works, at episode $k$, the agent rolls a trajectory and trains a policy before transitioning to episode $k+1$. In the context of the time-desynchronized environment, however, the agent at time $t_{k}$ allocates $Δt$ for trajectory generation and training, subsequently moves to the next episode at $t_{k+1}=t_{k}+Δt$. Despite a fixed total number of episodes ($K$), the agent accumulates different trajectories influenced by the choice of interaction times ($t_1,t_2,...,t_K$), significantly impacting the suboptimality gap of the policy. We propose a Proactively Synchronizing Tempo ($\texttt{ProST}$) framework that computes a suboptimal sequence {$t_1,t_2,...,t_K$} (= { $t_{1:K}$}) by minimizing an upper bound on its performance measure, i.e., the dynamic regret. Our main contribution is that we show that a suboptimal {$t_{1:K}$} trades-off between the policy training time (agent tempo) and how fast the environment changes (environment tempo). Theoretically, this work develops a suboptimal {$t_{1:K}$} as a function of the degree of the environment's non-stationarity while also achieving a sublinear dynamic regret. Our experimental evaluation on various high-dimensional non-stationary environments shows that the $\texttt{ProST}$ framework achieves a higher online return at suboptimal {$t_{1:K}$} than the existing methods.

Jongmin Lee, Yoonwoo Jeong, Minsu Cho

We study the problem of learning to assign a characteristic pose, i.e., scale and orientation, for an image region of interest. Despite its apparent simplicity, the problem is non-trivial; it is hard to obtain a large-scale set of image regions with explicit pose annotations that a model directly learns from. To tackle the issue, we propose a self-supervised learning framework with a histogram alignment technique. It generates pairs of image patches by random rescaling/rotating and then train an estimator to predict their scale/orientation values so that their relative difference is consistent with the rescaling/rotating used. The estimator learns to predict a non-parametric histogram distribution of scale/orientation without any supervision. Experiments show that it significantly outperforms previous methods in scale/orientation estimation and also improves image matching and 6 DoF camera pose estimation by incorporating our patch poses into a matching process.

Daiki E. Matsunaga, Jongmin Lee, Jaeseok Yoon et al.

One of the main challenges in offline Reinforcement Learning (RL) is the distribution shift that arises from the learned policy deviating from the data collection policy. This is often addressed by avoiding out-of-distribution (OOD) actions during policy improvement as their presence can lead to substantial performance degradation. This challenge is amplified in the offline Multi-Agent RL (MARL) setting since the joint action space grows exponentially with the number of agents. To avoid this curse of dimensionality, existing MARL methods adopt either value decomposition methods or fully decentralized training of individual agents. However, even when combined with standard conservatism principles, these methods can still result in the selection of OOD joint actions in offline MARL. To this end, we introduce AlberDICE, an offline MARL algorithm that alternatively performs centralized training of individual agents based on stationary distribution optimization. AlberDICE circumvents the exponential complexity of MARL by computing the best response of one agent at a time while effectively avoiding OOD joint action selection. Theoretically, we show that the alternating optimization procedure converges to Nash policies. In the experiments, we demonstrate that AlberDICE significantly outperforms baseline algorithms on a standard suite of MARL benchmarks.

Haanvid Lee, Jongmin Lee, Yunseon Choi et al.

We consider local kernel metric learning for off-policy evaluation (OPE) of deterministic policies in contextual bandits with continuous action spaces. Our work is motivated by practical scenarios where the target policy needs to be deterministic due to domain requirements, such as prescription of treatment dosage and duration in medicine. Although importance sampling (IS) provides a basic principle for OPE, it is ill-posed for the deterministic target policy with continuous actions. Our main idea is to relax the target policy and pose the problem as kernel-based estimation, where we learn the kernel metric in order to minimize the overall mean squared error (MSE). We present an analytic solution for the optimal metric, based on the analysis of bias and variance. Whereas prior work has been limited to scalar action spaces or kernel bandwidth selection, our work takes a step further being capable of vector action spaces and metric optimization. We show that our estimator is consistent, and significantly reduces the MSE compared to baseline OPE methods through experiments on various domains.

Jongmin Lee, Amin Rakhsha, Ernest K. Ryu et al.

The Value Iteration (VI) algorithm is an iterative procedure to compute the value function of a Markov decision process, and is the basis of many reinforcement learning (RL) algorithms as well. As the error convergence rate of VI as a function of iteration $k$ is $O(γ^k)$, it is slow when the discount factor $γ$ is close to $1$. To accelerate the computation of the value function, we propose Deflated Dynamics Value Iteration (DDVI). DDVI uses matrix splitting and matrix deflation techniques to effectively remove (deflate) the top $s$ dominant eigen-structure of the transition matrix $\mathcal{P}^π$. We prove that this leads to a $\tilde{O}(γ^k |λ_{s+1}|^k)$ convergence rate, where $λ_{s+1}$is $(s+1)$-th largest eigenvalue of the dynamics matrix. We then extend DDVI to the RL setting and present Deflated Dynamics Temporal Difference (DDTD) algorithm. We empirically show the effectiveness of the proposed algorithms.

Youngjoo Ahn, Sangyeop Yeo, Gijung Lim et al.

Automated proof generation for formal software verification remains largely unresolved despite advances in large language models (LLMs). While LLMs perform well in NLP, vision, and code generation, formal verification still requires substantial human effort. Interactive theorem proving (ITP) demands manual proof construction under strict logical constraints, limiting scalability; for example, verifying the seL4 microkernel required decades of effort. Existing LLM-based approaches attempt to automate this process but remain limited. Most rely on single-shot generation or shallow retrieval, which may work for small proofs but fail to scale to large, interdependent verification tasks with deep structural dependencies. We present PROMISE (PROof MIning via Structural Embeddings), a structure-aware framework that reframes proof generation as a stateful search over proof-state transitions. Instead of surface-level retrieval, PROMISE mines structural patterns from proof states and tactic transitions, enabling retrieval and adaptation of compatible proof fragments during iterative search. We evaluate PROMISE on the seL4 benchmark across multiple LLM backends and compare it with prior systems such as Selene and Rango. PROMISE consistently outperforms prior methods, achieving up to +26 point improvements (186% relative gain) while maintaining robustness across models, demonstrating the effectiveness of structure-aware proof mining for scalable theorem proving.

Yoon Jo Kim, Wonyoung Cho, Jongmin Lee et al.

Delineating the clinical target volume (CTV) in radiotherapy involves complex margins constrained by tumor location and anatomical barriers. While deep learning models automate this process, their rigid reliance on expert-annotated data requires costly retraining whenever clinical guidelines update. To overcome this limitation, we introduce OncoAgent, a novel guideline-aware AI agent framework that seamlessly converts textual clinical guidelines into three-dimensional target contours in a training-free manner. Evaluated on esophageal cancer cases, the agent achieves a zero-shot Dice similarity coefficient of 0.842 for the CTV and 0.880 for the planning target volume, demonstrating performance highly comparable to a fully supervised nnU-Net baseline. Notably, in a blinded clinical evaluation, physicians strongly preferred OncoAgent over the supervised baseline, rating it higher in guideline compliance, modification effort, and clinical acceptability. Furthermore, the framework generalizes zero-shot to alternative esophageal guidelines and other anatomical sites (e.g., prostate) without any retraining. Beyond mere volumetric overlap, our agent-based paradigm offers near-instantaneous adaptability to alternative guidelines, providing a scalable and transparent pathway toward interpretability in radiotherapy treatment planning.

Jongmin Lee, Seungyeop Kang, Sungjoo Yoo

Establishing consistent correspondences across images is essential for 3D vision tasks such as structure-from-motion (SfM), yet most existing matchers operate in a pairwise manner, often producing fragmented and geometrically inconsistent tracks when their predictions are chained across views. We propose MV-RoMa, a multi-view dense matching model that jointly estimates dense correspondences from a source image to multiple co-visible targets. Specifically, we design an efficient model architecture which avoids high computational cost of full cross-attention for multi-view feature interaction: (i) multi-view encoder that leverages pair-wise matching results as a geometric prior, and (ii) multi-view matching refiner that refines correspondences using pixel-wise attention. Additionally, we propose a post-processing strategy that integrates our model's consistent multi-view correspondences as high-quality tracks for SfM. Across diverse and challenging benchmarks, MV-RoMa produces more reliable correspondences and substantially denser, more accurate 3D reconstructions than existing sparse and dense matching methods. Project page: https://icetea-cv.github.io/mv-roma/.

Junwoo Chang, Minwoo Park, Joohwan Seo et al.

Group symmetries provide a powerful inductive bias for reinforcement learning (RL), enabling efficient generalization across symmetric states and actions via group-invariant Markov Decision Processes (MDPs). However, real-world environments almost never realize fully group-invariant MDPs; dynamics, actuation limits, and reward design usually break symmetries, often only locally. Under group-invariant Bellman backups for such cases, local symmetry-breaking introduces errors that propagate across the entire state-action space, resulting in global value estimation errors. To address this, we introduce Partially group-Invariant MDP (PI-MDP), which selectively applies group-invariant or standard Bellman backups depending on where symmetry holds. This framework mitigates error propagation from locally broken symmetries while maintaining the benefits of equivariance, thereby enhancing sample efficiency and generalizability. Building on this framework, we present practical RL algorithms -- Partially Equivariant (PE)-DQN for discrete control and PE-SAC for continuous control -- that combine the benefits of equivariance with robustness to symmetry-breaking. Experiments across Grid-World, locomotion, and manipulation benchmarks demonstrate that PE-DQN and PE-SAC significantly outperform baselines, highlighting the importance of selective symmetry exploitation for robust and sample-efficient RL.

Junwoo Chang, Joseph Park, Roberto Horowitz et al.

Unsupervised skill discovery aims to acquire behavior primitives that improve exploration and accelerate downstream task learning. However, existing approaches often ignore the geometric symmetries of physical environments, leading to redundant behaviors and sample inefficiency. To address this, we introduce Group-Invariant Skill Discovery (GISD), a framework that explicitly embeds group structure into the skill discovery objective. Our approach is grounded in a theoretical guarantee: we prove that in group-symmetric environments, the standard Wasserstein dependency measure admits a globally optimal solution comprised of an equivariant policy and a group-invariant scoring function. Motivated by this, we formulate the Group-Invariant Wasserstein dependency measure, which restricts the optimization to this symmetry-aware subspace without loss of optimality. Practically, we parameterize the scoring function using a group Fourier representation and define the intrinsic reward via the alignment of equivariant latent features, ensuring that the discovered skills generalize systematically under group transformations. Experiments on state-based and pixel-based locomotion benchmarks demonstrate that GISD achieves broader state-space coverage and improved efficiency in downstream task learning compared to a strong baseline.

Jongmin Lee, Meiqi Sun, Pieter Abbeel

In the unsupervised pre-training for reinforcement learning, the agent aims to learn a prior policy for downstream tasks without relying on task-specific reward functions. We focus on state entropy maximization (SEM), where the goal is to learn a policy that maximizes the entropy of the state stationary distribution. In this paper, we introduce SEMDICE, a principled off-policy algorithm that computes an SEM policy from an arbitrary off-policy dataset, which optimizes the policy directly within the space of stationary distributions. SEMDICE computes a single, stationary Markov state-entropy-maximizing policy from an arbitrary off-policy dataset. Experimental results demonstrate that SEMDICE outperforms baseline algorithms in maximizing state entropy while achieving the best adaptation efficiency for downstream tasks among SEM-based unsupervised RL pre-training methods.

JongMin Lee, Sungjoo Yoo

We present Dense-SfM, a novel Structure from Motion (SfM) framework designed for dense and accurate 3D reconstruction from multi-view images. Sparse keypoint matching, which traditional SfM methods often rely on, limits both accuracy and point density, especially in texture-less areas. Dense-SfM addresses this limitation by integrating dense matching with a Gaussian Splatting (GS) based track extension which gives more consistent, longer feature tracks. To further improve reconstruction accuracy, Dense-SfM is equipped with a multi-view kernelized matching module leveraging transformer and Gaussian Process architectures, for robust track refinement across multi-views. Evaluations on the ETH3D and Texture-Poor SfM datasets show that Dense-SfM offers significant improvements in accuracy and density over state-of-the-art methods. Project page: https://icetea-cv.github.io/densesfm/.

Lennart Bamberg, Filippo Minnella, Roberto Bosio et al.

Neural Processing Units (NPUs) are key to enabling efficient AI inference in resource-constrained edge environments. While peak tera operations per second (TOPS) is often used to gauge performance, it poorly reflects real-world performance and typically rather correlates with higher silicon cost. To address this, architects must focus on maximizing compute utilization, without sacrificing flexibility. This paper presents the eIQ Neutron efficient-NPU, integrated into a commercial flagship MPU, alongside co-designed compiler algorithms. The architecture employs a flexible, data-driven design, while the compiler uses a constrained programming approach to optimize compute and data movement based on workload characteristics. Compared to the leading embedded NPU and compiler stack, our solution achieves an average speedup of 1.8x (4x peak) at equal TOPS and memory resources across standard AI-benchmarks. Even against NPUs with double the compute and memory resources, Neutron delivers up to 3.3x higher performance.

Jongmin Lee, Minsu Cho

Determining the 3D orientations of an object in an image, known as single-image pose estimation, is a crucial task in 3D vision applications. Existing methods typically learn 3D rotations parametrized in the spatial domain using Euler angles or quaternions, but these representations often introduce discontinuities and singularities. SO(3)-equivariant networks enable the structured capture of pose patterns with data-efficient learning, but the parametrizations in spatial domain are incompatible with their architecture, particularly spherical CNNs, which operate in the frequency domain to enhance computational efficiency. To overcome these issues, we propose a frequency-domain approach that directly predicts Wigner-D coefficients for 3D rotation regression, aligning with the operations of spherical CNNs. Our SO(3)-equivariant pose harmonics predictor overcomes the limitations of spatial parameterizations, ensuring consistent pose estimation under arbitrary rotations. Trained with a frequency-domain regression loss, our method achieves state-of-the-art results on benchmarks such as ModelNet10-SO(3) and PASCAL3D+, with significant improvements in accuracy, robustness, and data efficiency.

Sang Hun Kim, Jongmin Lee, Dongkyu Park et al.

Human consciousness is still a concept hard to define with current scientific understanding. Although Large Language Models (LLMs) have recently demonstrated significant advancements across various domains including translation and summarization, human consciousness is not something to imitate with current upfront technology owing to so-called hallucination. This study, therefore, proposes a novel approach to address these challenges by integrating psychoanalysis and the Myers-Briggs Type Indicator (MBTI) into constructing consciousness and personality modules. We developed three artificial consciousnesses (self-awareness, unconsciousness, and preconsciousness) based on the principles of psychoanalysis. Additionally, we designed 16 characters with different personalities representing the sixteen MBTI types, with several attributes such as needs, status, and memories. To determine if our model's artificial consciousness exhibits human-like cognition, we created ten distinct situations considering seven attributes such as emotional understanding and logical thinking. The decision-making process of artificial consciousness and the final action were evaluated in three ways: survey evaluation, three-tier classification via ChatGPT, and qualitative review. Both quantitative and qualitative analyses indicated a high likelihood of well-simulated consciousness, although the difference in response between different characters and consciousnesses was not very significant. This implies that the developed models incorporating elements of psychoanalysis and personality theory can lead to building a more intuitive and adaptable AI system with humanoid consciousness. Therefore, this study contributes to opening up new avenues for improving AI interactions in complex cognitive contexts.

Jongmin Lee, Ernest K. Ryu

The classical policy gradient method is the theoretical and conceptual foundation of modern policy-based reinforcement learning (RL) algorithms. Most rigorous analyses of such methods, particularly those establishing convergence guarantees, assume a discount factor $γ< 1$. In contrast, however, a recent line of work on policy-based RL for large language models uses the undiscounted total-reward setting with $γ= 1$, rendering much of the existing theory inapplicable. In this paper, we provide analyses of the policy gradient method for undiscounted expected total-reward infinite-horizon MDPs based on two key insights: (i) the classification of the MDP states into recurrent and transient states is invariant over the set of policies that assign strictly positive probability to every action (as is typical in deep RL models employing a softmax output layer) and (ii) the classical state visitation measure (which may be ill-defined when $γ= 1$) can be replaced with a new object that we call the transient visitation measure.

Jongmin Lee, Ernest K. Ryu

Although there is an extensive body of work characterizing the sample complexity of discounted-return offline RL with function approximations, prior work on the average-reward setting has received significantly less attention, and existing approaches rely on restrictive assumptions, such as ergodicity or linearity of the MDP. In this work, we establish the first sample complexity results for average-reward offline RL with function approximation for weakly communicating MDPs, a much milder assumption. To this end, we introduce Anchored Fitted Q-Iteration, which combines the standard Fitted Q-Iteration with an anchor mechanism. We show that the anchor, which can be interpreted as a form of weight decay, is crucial for enabling finite-time analysis in the average-reward setting. We also extend our finite-time analysis to the setup where the dataset is generated from a single-trajectory rather than IID transitions, again leveraging the anchor mechanism.

Sang Hun Kim, Jongmin Lee, Dongkyu Park et al.

We propose a multi-agent framework for modeling artificial consciousness in large language models (LLMs), grounded in psychoanalytic theory. Our \textbf{Psychodynamic Model} simulates self-awareness, preconsciousness, and unconsciousness through agent interaction, guided by a Personalization Module combining fixed traits and dynamic needs. Using parameter-efficient fine-tuning on emotionally rich dialogues, the system was evaluated across eight personalized conditions. An LLM as a judge approach showed a 71.2\% preference for the fine-tuned model, with improved emotional depth and reduced output variance, demonstrating its potential for adaptive, personalized cognition.

Woosung Kim, JunHo Seo, Jongmin Lee et al.

Stationary Distribution Correction Estimation (DICE) addresses the mismatch between the stationary distribution induced by a policy and the target distribution required for reliable off-policy evaluation (OPE) and policy optimization. DICE-based offline constrained RL particularly benefits from the flexibility of DICE, as it simultaneously maximizes return while estimating costs in offline settings. However, we have observed that recent approaches designed to enhance the offline RL performance of the DICE framework inadvertently undermine its ability to perform OPE, making them unsuitable for constrained RL scenarios. In this paper, we identify the root cause of this limitation: their reliance on a semi-gradient optimization, which solves a fundamentally different optimization problem and results in failures in cost estimation. Building on these insights, we propose a novel method to enable OPE and constrained RL through semi-gradient DICE. Our method ensures accurate cost estimation and achieves state-of-the-art performance on the offline constrained RL benchmark, DSRL.

Woosung Kim, Jinho Lee, Jongmin Lee et al.

Multi-objective reinforcement learning (MORL) aims to optimize policies in the presence of conflicting objectives, where linear scalarization is commonly used to reduce vector-valued returns into scalar signals. While effective for certain preferences, this approach cannot capture fairness-oriented goals such as Nash social welfare or max-min fairness, which require nonlinear and non-additive trade-offs. Although several online algorithms have been proposed for specific fairness objectives, a unified approach for optimizing nonlinear welfare criteria in the offline setting-where learning must proceed from a fixed dataset-remains unexplored. In this work, we present FairDICE, the first offline MORL framework that directly optimizes nonlinear welfare objective. FairDICE leverages distribution correction estimation to jointly account for welfare maximization and distributional regularization, enabling stable and sample-efficient learning without requiring explicit preference weights or exhaustive weight search. Across multiple offline benchmarks, FairDICE demonstrates strong fairness-aware performance compared to existing baselines.

LG AI Research, Sehyun Chun, Jiye Kim et al.

The extraction of molecular structures and reaction data from scientific documents is challenging due to their varied, unstructured chemical formats and complex document layouts. To address this, we introduce MolMole, a vision-based deep learning framework that unifies molecule detection, reaction diagram parsing, and optical chemical structure recognition (OCSR) into a single pipeline for automating the extraction of chemical data directly from page-level documents. Recognizing the lack of a standard page-level benchmark and evaluation metric, we also present a testset of 550 pages annotated with molecule bounding boxes, reaction labels, and MOLfiles, along with a novel evaluation metric. Experimental results demonstrate that MolMole outperforms existing toolkits on both our benchmark and public datasets. The benchmark testset will be publicly available, and the MolMole toolkit will be accessible soon through an interactive demo on the LG AI Research website. For commercial inquiries, please contact us at \href{mailto:contact_ddu@lgresearch.ai}{contact\_ddu@lgresearch.ai}.

Jongmin Lee, Ernest K. Ryu

Value Iteration (VI) is foundational to the theory and practice of modern reinforcement learning, and it is known to converge at a $\mathcal{O}(γ^k)$-rate, where $γ$ is the discount factor. Surprisingly, however, the optimal rate for the VI setup was not known, and finding a general acceleration mechanism has been an open problem. In this paper, we present the first accelerated VI for both the Bellman consistency and optimality operators. Our method, called Anc-VI, is based on an \emph{anchoring} mechanism (distinct from Nesterov's acceleration), and it reduces the Bellman error faster than standard VI. In particular, Anc-VI exhibits a $\mathcal{O}(1/k)$-rate for $γ\approx 1$ or even $γ=1$, while standard VI has rate $\mathcal{O}(1)$ for $γ\ge 1-1/k$, where $k$ is the iteration count. We also provide a complexity lower bound matching the upper bound up to a constant factor of $4$, thereby establishing optimality of the accelerated rate of Anc-VI. Finally, we show that the anchoring mechanism provides the same benefit in the approximate VI and Gauss--Seidel VI setups as well.

Geon-Hyeong Kim, Jongmin Lee, Youngsoo Jang et al.

We consider the problem of learning from observation (LfO), in which the agent aims to mimic the expert's behavior from the state-only demonstrations by experts. We additionally assume that the agent cannot interact with the environment but has access to the action-labeled transition data collected by some agents with unknown qualities. This offline setting for LfO is appealing in many real-world scenarios where the ground-truth expert actions are inaccessible and the arbitrary environment interactions are costly or risky. In this paper, we present LobsDICE, an offline LfO algorithm that learns to imitate the expert policy via optimization in the space of stationary distributions. Our algorithm solves a single convex minimization problem, which minimizes the divergence between the two state-transition distributions induced by the expert and the agent policy. Through an extensive set of offline LfO tasks, we show that LobsDICE outperforms strong baseline methods.

Jongmin Lee, Joo Young Choi, Ernest K. Ryu et al.

The tremendous recent progress in analyzing the training dynamics of overparameterized neural networks has primarily focused on wide networks and therefore does not sufficiently address the role of depth in deep learning. In this work, we present the first trainability guarantee of infinitely deep but narrow neural networks. We study the infinite-depth limit of a multilayer perceptron (MLP) with a specific initialization and establish a trainability guarantee using the NTK theory. We then extend the analysis to an infinitely deep convolutional neural network (CNN) and perform brief experiments.

Jongmin Lee, Wonseok Jeon, Byung-Jun Lee et al.

We consider the offline reinforcement learning (RL) setting where the agent aims to optimize the policy solely from the data without further environment interactions. In offline RL, the distributional shift becomes the primary source of difficulty, which arises from the deviation of the target policy being optimized from the behavior policy used for data collection. This typically causes overestimation of action values, which poses severe problems for model-free algorithms that use bootstrapping. To mitigate the problem, prior offline RL algorithms often used sophisticated techniques that encourage underestimation of action values, which introduces an additional set of hyperparameters that need to be tuned properly. In this paper, we present an offline RL algorithm that prevents overestimation in a more principled way. Our algorithm, OptiDICE, directly estimates the stationary distribution corrections of the optimal policy and does not rely on policy-gradients, unlike previous offline RL algorithms. Using an extensive set of benchmark datasets for offline RL, we show that OptiDICE performs competitively with the state-of-the-art methods.

Juhong Min, Jongmin Lee, Jean Ponce et al.

Feature representation plays a crucial role in visual correspondence, and recent methods for image matching resort to deeply stacked convolutional layers. These models, however, are both monolithic and static in the sense that they typically use a specific level of features, e.g., the output of the last layer, and adhere to it regardless of the images to match. In this work, we introduce a novel approach to visual correspondence that dynamically composes effective features by leveraging relevant layers conditioned on the images to match. Inspired by both multi-layer feature composition in object detection and adaptive inference architectures in classification, the proposed method, dubbed Dynamic Hyperpixel Flow, learns to compose hypercolumn features on the fly by selecting a small number of relevant layers from a deep convolutional neural network. We demonstrate the effectiveness on the task of semantic correspondence, i.e., establishing correspondences between images depicting different instances of the same object or scene category. Experiments on standard benchmarks show that the proposed method greatly improves matching performance over the state of the art in an adaptive and efficient manner.

Jang-Hyun Kim, Jongmin Lee, Hee-Seok Oh

This paper presents a new approach for dimension reduction of data observed in a sphere. Several dimension reduction techniques have recently developed for the analysis of non-Euclidean data. As a pioneer work, Hauberg (2016) attempted to implement principal curves on Riemannian manifolds. However, this approach uses approximations to deal with data on Riemannian manifolds, which causes distorted results. In this study, we propose a new approach to construct principal curves on a sphere by a projection of the data onto a continuous curve. Our approach lies in the same line of Hastie and Stuetzle (1989) that proposed principal curves for Euclidean space data. We further investigate the stationarity of the proposed principal curves that satisfy the self-consistency on a sphere. Results from real data analysis with earthquake data and simulation examples demonstrate the promising empirical properties of the proposed approach.

Juhong Min, Jongmin Lee, Jean Ponce et al.

Establishing visual correspondences under large intra-class variations, which is often referred to as semantic correspondence or semantic matching, remains a challenging problem in computer vision. Despite its significance, however, most of the datasets for semantic correspondence are limited to a small amount of image pairs with similar viewpoints and scales. In this paper, we present a new large-scale benchmark dataset of semantically paired images, SPair-71k, which contains 70,958 image pairs with diverse variations in viewpoint and scale. Compared to previous datasets, it is significantly larger in number and contains more accurate and richer annotations. We believe this dataset will provide a reliable testbed to study the problem of semantic correspondence and will help to advance research in this area. We provide the results of recent methods on our new dataset as baselines for further research. Our benchmark is available online at http://cvlab.postech.ac.kr/research/SPair-71k/.

Juhong Min, Jongmin Lee, Jean Ponce et al.

Establishing visual correspondences under large intra-class variations requires analyzing images at different levels, from features linked to semantics and context to local patterns, while being invariant to instance-specific details. To tackle these challenges, we represent images by "hyperpixels" that leverage a small number of relevant features selected among early to late layers of a convolutional neural network. Taking advantage of the condensed features of hyperpixels, we develop an effective real-time matching algorithm based on Hough geometric voting. The proposed method, hyperpixel flow, sets a new state of the art on three standard benchmarks as well as a new dataset, SPair-71k, which contains a significantly larger number of image pairs than existing datasets, with more accurate and richer annotations for in-depth analysis.

Paul Hongsuck Seo, Jongmin Lee, Deunsol Jung et al.

Semantic correspondence is the problem of establishing correspondences across images depicting different instances of the same object or scene class. One of recent approaches to this problem is to estimate parameters of a global transformation model that densely aligns one image to the other. Since an entire correlation map between all feature pairs across images is typically used to predict such a global transformation, noisy features from different backgrounds, clutter, and occlusion distract the predictor from correct estimation of the alignment. This is a challenging issue, in particular, in the problem of semantic correspondence where a large degree of image variations is often involved. In this paper, we introduce an attentive semantic alignment method that focuses on reliable correlations, filtering out distractors. For effective attention, we also propose an offset-aware correlation kernel that learns to capture translation-invariant local transformations in computing correlation values over spatial locations. Experiments demonstrate the effectiveness of the attentive model and offset-aware kernel, and the proposed model combining both techniques achieves the state-of-the-art performance.