arXiv subjects

Masashi Sugiyama

Publications and source records attributed to Masashi Sugiyama.

At least 55 records · Page 3Linked to original sources

arXiv2026

Bifrost: Steering Strategic Trajectories to Bridge Contextual Gaps for Self-Improving Agents

Autonomous agents excel in self-improvement through reflection and iterative refinement, which reuse successful task trajectories as in-context examples to assist subsequent reasoning. However, shifting across tasks often introduces a context mismatch. Hence, existing approaches either discard the trajectories or manipulate them using heuristics, leading to a non-negligible fine-tuning cost or unguaranteed performance. To bridge this gap, we reveal a context-trajectory correlation, where shifts of context are highly parallel with shifts of trajectory. Based on this finding, we propose BrIdge contextual gap FoR imprOvised trajectory STeering (Bifrost), a training-free method that leverages context differences to precisely guide the adaptation of previously solved trajectories towards the target task, mitigating the misalignment caused by context shifts. Our trajectory adaptation is conducted at the representation level using agent hidden states, ensuring trajectory transformation accurately aligns with the target context in a shared space. Across diverse benchmarks, Bifrost consistently outperforms existing trajectory reuse and finetuned self-improvement methods, demonstrating that agents can effectively leverage past experiences despite substantial context shifts.

cs.LG↗

arXiv2026

Causal Graph Learning via Distributional Invariance of Cause-Effect Relationship

This paper introduces a new framework for recovering causal graphs from observational data, leveraging the observation that the distribution of an effect, conditioned on its causes, remains invariant to changes in the prior distribution of those causes. This insight enables a direct test for potential causal relationships by checking the variance of their corresponding effect-cause conditional distributions across multiple downsampled subsets of the data. These subsets are selected to reflect different prior cause distributions, while preserving the effect-cause conditional relationships. Using this invariance test and exploiting an (empirical) sparsity of most causal graphs, we develop an algorithm that efficiently uncovers causal relationships with quadratic complexity in the number of observational variables, reducing the processing time by up to 25x compared to state-of-the-art methods. Our empirical experiments on a varied benchmark of large-scale datasets show superior or equivalent performance compared to existing works, while achieving enhanced scalability.

cs.LG↗

arXiv2026

Positive-Unlabeled Reinforcement Learning Distillation for On-Premise Small Models

Due to constraints on privacy, cost, and latency, on-premise deployment of small models is increasingly common. However, most practical pipelines stop at supervised fine-tuning (SFT) and fail to reach the reinforcement learning (RL) alignment stage. The main reason is that RL alignment typically requires either expensive human preference annotation or heavy reliance on high-quality reward models with large-scale API usage and ongoing engineering maintenance, both of which are ill-suited to on-premise settings. To bridge this gap, we propose a positive-unlabeled (PU) RL distillation method for on-premise small-model deployment. Without human-labeled preferences or a reward model, our method distills the teacher's preference-optimization capability from black-box generations into a locally trainable student. For each prompt, we query the teacher once to obtain an anchor response, locally sample multiple student candidates, and perform anchor-conditioned self-ranking to induce pairwise or listwise preferences, enabling a fully local training loop via direct preference optimization or group relative policy optimization. Theoretical analysis justifies that the induced preference signal by our method is order-consistent and concentrates on near-optimal candidates, supporting its stability for preference optimization. Experiments demonstrate that our method achieves consistently strong performance under a low-cost setting.

cs.LG↗

arXiv2026

HapticMatch: An Exploration for Generative Material Haptic Simulation and Interaction

High-fidelity haptic feedback is essential for immersive virtual environments, yet authoring realistic tactile textures remains a significant bottleneck for designers. We introduce HapticMatch, a visual-to-tactile generation framework designed to democratize haptic content creation. We present a novel dataset containing precisely aligned pairs of micro-scale optical images, surface height maps, and friction-induced vibrations for 100 diverse materials. Leveraging this data, we explore and demonstrate that conditional generative models like diffusion and flow-matching can synthesize high-fidelity, renderable surface geometries directly from standard RGB photos. By enabling a "Scan-to-Touch" workflow, HapticMatch allows interaction designers to rapidly prototype multimodal surface sensations without specialized recording equipment, bridging the gap between visual and tactile immersion in VR/AR interfaces.

cs.HC↗

arXiv2025

From Coefficients to Directions: Rethinking Model Merging with Directional Alignment

Model merging has emerged as a practical paradigm for integrating multiple independently trained models into a single model without joint retraining. Previous studies have demonstrated the effectiveness of combining parameters through strategies such as parameter decomposition, coefficient optimization, and subspace learning, significantly reducing the need for expensive joint training and achieving strong empirical performance across diverse tasks. However, these approaches predominantly treat merging as a problem of parameter space decomposition or fusion coefficient optimization, while overlooking the critical role of directional information in both parameter and feature spaces. In practice, naïve merging introduces inconsistencies in dominant parameter directions and disrupts structural coherence across models, which can degrade performance. Moreover, coefficient-based optimization methods implicitly assume compatible feature-space directions across models. However, Neural Collapse indicates that class features follow structured directional patterns, which may differ across independently trained models, making coefficient optimization alone insufficient. In this work, we emphasize the importance of \emph{directional alignment} and introduce a unified geometric framework, \emph{Merging with Directional Alignment} (\method{}), which aligns directional structures consistently in both the parameter and feature spaces. Our analysis shows that directional alignment improves structural coherence, and extensive experiments across benchmarks, model scales, and task configurations further validate the effectiveness of our approach.

cs.LG↗

arXiv2025

Generalized Linear Bandits: Almost Optimal Regret with One-Pass Update

We study the generalized linear bandit (GLB) problem, a contextual multi-armed bandit framework that extends the classical linear model by incorporating a non-linear link function, thereby modeling a broad class of reward distributions such as Bernoulli and Poisson. While GLBs are widely applicable to real-world scenarios, their non-linear nature introduces significant challenges in achieving both computational and statistical efficiency. Existing methods typically trade off between two objectives, either incurring high per-round costs for optimal regret guarantees or compromising statistical efficiency to enable constant-time updates. In this paper, we propose a jointly efficient algorithm that attains a nearly optimal regret bound with $\mathcal{O}(1)$ time and space complexities per round. The core of our method is a tight confidence set for the online mirror descent (OMD) estimator, which is derived through a novel analysis that leverages the notion of mix loss from online prediction. The analysis shows that our OMD estimator, even with its one-pass updates, achieves statistical efficiency comparable to maximum likelihood estimation, thereby leading to a jointly efficient optimistic method.

cs.LG↗

arXiv2025

Rethinking Toxicity Evaluation in Large Language Models: A Multi-Label Perspective

Large language models (LLMs) have achieved impressive results across a range of natural language processing tasks, but their potential to generate harmful content has raised serious safety concerns. Current toxicity detectors primarily rely on single-label benchmarks, which cannot adequately capture the inherently ambiguous and multi-dimensional nature of real-world toxic prompts. This limitation results in biased evaluations, including missed toxic detections and false positives, undermining the reliability of existing detectors. Additionally, gathering comprehensive multi-label annotations across fine-grained toxicity categories is prohibitively costly, further hindering effective evaluation and development. To tackle these issues, we introduce three novel multi-label benchmarks for toxicity detection: \textbf{Q-A-MLL}, \textbf{R-A-MLL}, and \textbf{H-X-MLL}, derived from public toxicity datasets and annotated according to a detailed 15-category taxonomy. We further provide a theoretical proof that, on our released datasets, training with pseudo-labels yields better performance than directly learning from single-label supervision. In addition, we develop a pseudo-label-based toxicity detection method. Extensive experimental results show that our approach significantly surpasses advanced baselines, including GPT-4o and DeepSeek, thus enabling more accurate and reliable evaluation of multi-label toxicity in LLM-generated content.

cs.CL↗

arXiv2025

Learning Robust Diffusion Models from Imprecise Supervision

Conditional diffusion models have achieved remarkable success in various generative tasks recently, but their training typically relies on large-scale datasets that inevitably contain imprecise information in conditional inputs. Such supervision, often stemming from noisy, ambiguous, or incomplete labels, will cause condition mismatch and degrade generation quality. To address this challenge, we propose DMIS, a unified framework for training robust Diffusion Models from Imprecise Supervision, which is the first systematic study within diffusion models. Our framework is derived from likelihood maximization and decomposes the objective into generative and classification components: the generative component models imprecise-label distributions, while the classification component leverages a diffusion classifier to infer class-posterior probabilities, with its efficiency further improved by an optimized timestep sampling strategy. Extensive experiments on diverse forms of imprecise supervision, covering tasks of image generation, weakly supervised learning, and noisy dataset condensation demonstrate that DMIS consistently produces high-quality and class-discriminative samples.

cs.LG↗

arXiv2025

LLM Routing with Dueling Feedback

We study LLM routing, the problem of selecting the best model for each query while balancing user satisfaction, model expertise, and inference cost. We formulate routing as contextual dueling bandits, learning from pairwise preference feedback rather than absolute scores, thereby yielding label-efficient and dynamic adaptation. Building on this formulation, we introduce Category-Calibrated Fine-Tuning (CCFT), a representation-learning method that derives model embeddings from offline data using contrastive fine-tuning with categorical weighting. These embeddings enable the practical instantiation of Feel-Good Thompson Sampling for Contextual Dueling Bandits (FGTS.CDB), a theoretically grounded posterior-sampling algorithm. We propose four variants of the categorical weighting that explicitly integrate model quality and cost, and we empirically evaluate the proposed methods on the RouterBench and MixInstruct datasets. Across both benchmarks, our methods achieve lower cumulative regret and faster convergence, with better robustness and performance-cost balance than strong baselines built with a general-purpose OpenAI embedding model.

cs.LG↗

arXiv2025

Parallel Simulation for Log-concave Sampling and Score-based Diffusion Models

Sampling from high-dimensional probability distributions is fundamental in machine learning and statistics. As datasets grow larger, computational efficiency becomes increasingly important, particularly in reducing adaptive complexity, namely the number of sequential rounds required for sampling algorithms. While recent works have introduced several parallelizable techniques, they often exhibit suboptimal convergence rates and remain significantly weaker than the latest lower bounds for log-concave sampling. To address this, we propose a novel parallel sampling method that improves adaptive complexity dependence on dimension $d$ reducing it from $\widetilde{\mathcal{O}}(\log^2 d)$ to $\widetilde{\mathcal{O}}(\log d)$. which is even optimal for log-concave sampling with some specific adaptive complexity. Our approach builds on parallel simulation techniques from scientific computing.

cs.DS↗

arXiv2025

Recursive Reward Aggregation

In reinforcement learning (RL), aligning agent behavior with specific objectives typically requires careful design of the reward function, which can be challenging when the desired objectives are complex. In this work, we propose an alternative approach for flexible behavior alignment that eliminates the need to modify the reward function by selecting appropriate reward aggregation functions. By introducing an algebraic perspective on Markov decision processes (MDPs), we show that the Bellman equations naturally emerge from the recursive generation and aggregation of rewards, allowing for the generalization of the standard discounted sum to other recursive aggregations, such as discounted max and Sharpe ratio. Our approach applies to both deterministic and stochastic settings and integrates seamlessly with value-based and actor-critic algorithms. Experimental results demonstrate that our approach effectively optimizes diverse objectives, highlighting its versatility and potential for real-world applications.

cs.LG↗

arXiv2025

Thompson Exploration with Best Challenger Rule in Best Arm Identification

This paper studies the fixed-confidence best arm identification (BAI) problem in the bandit framework in the canonical single-parameter exponential models. For this problem, many policies have been proposed, but most of them require solving an optimization problem at every round and/or are forced to explore an arm at least a certain number of times except those restricted to the Gaussian model. To address these limitations, we propose a novel policy that combines Thompson sampling with a computationally efficient approach known as the best challenger rule. While Thompson sampling was originally considered for maximizing the cumulative reward, we demonstrate that it can be used to naturally explore arms in BAI without forcing it. We show that our policy is asymptotically optimal for any two-armed bandit problems and achieves near optimality for general $K$-armed bandit problems for $K\geq 3$. Nevertheless, in numerical experiments, our policy shows competitive performance compared to asymptotically optimal policies in terms of sample complexity while requiring less computation cost. In addition, we highlight the advantages of our policy by comparing it to the concept of $β$-optimality, a relaxed notion of asymptotic optimality commonly considered in the analysis of a class of policies including the proposed one.

stat.ML↗

arXiv2025

What Makes "Good" Distractors for Object Hallucination Evaluation in Large Vision-Language Models?

Large Vision-Language Models (LVLMs), empowered by the success of Large Language Models (LLMs), have achieved impressive performance across domains. Despite the great advances in LVLMs, they still suffer from the unavailable object hallucination issue, which tends to generate objects inconsistent with the image content. The most commonly used Polling-based Object Probing Evaluation (POPE) benchmark evaluates this issue by sampling negative categories according to category-level statistics, \textit{e.g.}, category frequencies and co-occurrence. However, with the continuous advancement of LVLMs, the POPE benchmark has shown diminishing effectiveness in assessing object hallucination, as it employs a simplistic sampling strategy that overlooks image-specific information and restricts distractors to negative object categories only. In this paper, we introduce the Hallucination searching-based Object Probing Evaluation (HOPE) benchmark, aiming to generate the most misleading distractors (\textit{i.e.}, non-existent objects or incorrect image descriptions) that can trigger hallucination in LVLMs, which serves as a means to more rigorously assess their immunity to hallucination. To explore the image-specific information, the content-aware hallucination searching leverages Contrastive Language-Image Pre-Training (CLIP) to approximate the predictive behavior of LVLMs by selecting negative objects with the highest predicted likelihood as distractors. To expand the scope of hallucination assessment, the description-based hallucination searching constructs highly misleading distractors by pairing true objects with false descriptions. Experimental results show that HOPE leads to a precision drop of at least 9\% and up to 23\% across various state-of-the-art LVLMs, significantly outperforming POPE in exposing hallucination vulnerabilities. The code is available at https://github.com/xiemk/HOPE.

cs.CV↗

arXiv2025

Robust Multi-View Learning via Representation Fusion of Sample-Level Attention and Alignment of Simulated Perturbation

Recently, multi-view learning (MVL) has garnered significant attention due to its ability to fuse discriminative information from multiple views. However, real-world multi-view datasets are often heterogeneous and imperfect, which usually causes MVL methods designed for specific combinations of views to lack application potential and limits their effectiveness. To address this issue, we propose a novel robust MVL method (namely RML) with simultaneous representation fusion and alignment. Specifically, we introduce a simple yet effective multi-view transformer fusion network where we transform heterogeneous multi-view data into homogeneous word embeddings, and then integrate multiple views by the sample-level attention mechanism to obtain a fused representation. Furthermore, we propose a simulated perturbation based multi-view contrastive learning framework that dynamically generates the noise and unusable perturbations for simulating imperfect data conditions. The simulated noisy and unusable data obtain two distinct fused representations, and we utilize contrastive learning to align them for learning discriminative and robust representations. Our RML is self-supervised and can also be applied for downstream tasks as a regularization. In experiments, we employ it in multi-view unsupervised clustering, noise-label classification, and as a plug-and-play module for cross-modal hashing retrieval. Extensive comparison experiments and ablation studies validate RML's effectiveness. Code is available at https://github.com/SubmissionsIn/RML.

cs.CV↗

arXiv2025

Off-Policy Corrected Reward Modeling for Reinforcement Learning from Human Feedback

Reinforcement Learning from Human Feedback (RLHF) allows us to train models, such as language models (LMs), to follow complex human preferences. In RLHF for LMs, we first train an LM using supervised fine-tuning, sample pairs of responses, obtain human feedback, and use the resulting data to train a reward model (RM). RL methods are then used to train the LM to maximize the reward given by the RM. As training progresses, the responses generated by the LM no longer resemble the responses seen by the RM during training, leading to the RM becoming inaccurate. The score given by the RM keeps increasing, but the learned behavior no longer matches the human preferences. This issue is known as overoptimization. We investigate overoptimization from the point of view of distribution shift and show that the shift results in an inconsistent estimate of the RM parameters, leading to an inconsistent estimate of the policy gradient. We propose Off-Policy Corrected Reward Modeling (OCRM), which iteratively off-policy corrects the RM using importance weighting, without requiring new labels or samples. This results in a more accurate RM, which empirically leads to an improved final policy. We validate our approach in experiments with summarization and chatbot datasets and show that it performs significantly better than standard RLHF methods and baselines. Our implementation is available at https://github.com/JohannesAck/OffPolicyCorrectedRewardModeling

cs.LG↗

arXiv2025

Non-stationary Online Learning for Curved Losses: Improved Dynamic Regret via Mixability

Non-stationary online learning has drawn much attention in recent years. Despite considerable progress, dynamic regret minimization has primarily focused on convex functions, leaving the functions with stronger curvature (e.g., squared or logistic loss) underexplored. In this work, we address this gap by showing that the regret can be substantially improved by leveraging the concept of mixability, a property that generalizes exp-concavity to effectively capture loss curvature. Let $d$ denote the dimensionality and $P_T$ the path length of comparators that reflects the environmental non-stationarity. We demonstrate that an exponential-weight method with fixed-share updates achieves an $\mathcal{O}(d T^{1/3} P_T^{2/3} \log T)$ dynamic regret for mixable losses, improving upon the best-known $\mathcal{O}(d^{10/3} T^{1/3} P_T^{2/3} \log T)$ result (Baby and Wang, 2021) in $d$. More importantly, this improvement arises from a simple yet powerful analytical framework that exploits the mixability, which avoids the Karush-Kuhn-Tucker-based analysis required by existing work.

cs.LG↗

arXiv2025

On Symmetric Losses for Robust Policy Optimization with Noisy Preferences

Optimizing policies based on human preferences is key to aligning language models with human intent. This work focuses on reward modeling, a core component in reinforcement learning from human feedback (RLHF), and offline preference optimization, such as direct preference optimization. Conventional approaches typically assume accurate annotations. However, real-world preference data often contains noise due to human errors or biases. We propose a principled framework for robust policy optimization under noisy preferences, viewing reward modeling as a classification problem. This allows us to leverage symmetric losses, known for their robustness to label noise in classification, leading to our Symmetric Preference Optimization (SymPO) method. We prove that symmetric losses enable successful policy optimization even under noisy labels, as the resulting reward remains rank-preserving -- a property sufficient for policy improvement. Experiments on synthetic and real-world tasks demonstrate the effectiveness of SymPO.

cs.LG↗

arXiv2025

In-context Demonstration Matters: On Prompt Optimization for Pseudo-Supervision Refinement

Large language models (LLMs) have achieved great success across diverse tasks, and fine-tuning is sometimes needed to further enhance generation quality. Most existing methods rely on human supervision or parameter retraining, both of which are costly in terms of data collection and computational resources. To handle these challenges, a direct solution is to generate ``high-confidence'' data from unsupervised downstream tasks and use them for in-context prompting or prompt optimization to refine the pseudo-supervision. However, relying solely on such data may lead to overfitting. In this paper, we leverage the in-context learning (ICL) abilities of LLMs and propose a novel approach, pseudo-supervised demonstrations aligned prompt optimization (PAPO) algorithm, which jointly refines both the prompt and the overall pseudo-supervision. The proposed learning objective ensures that the optimized prompt guides the LLM to generate consistent responses for a given input when pseudo-supervised data from the downstream task are used as demonstrations, enabling refinement over the entire pseudo-supervision. The prompt is optimized by translating gradient signals into textual critiques, which serve as feedback to iteratively refine the prompt and model responses. Theoretical analysis in a simplified classification setting shows that the refined pseudo-supervision exhibits a geometric clustering structure, helping to mitigate overfitting. Experiments on question answering, natural language inference benchmarks, and a real-world molecule optimization task, show the effectiveness of the proposed algorithm.

cs.CL↗