Hikaru Asano

Hikaru Asano, Tadashi Kozuno, Kuniaki Saito et al.

Masked Diffusion Language Models (MDLMs) generate text by iteratively filling masked tokens, requiring two coupled decisions at each step: which positions to unmask (where-to-unmask) and which tokens to place (what-to-unmask). While standard MDLM training directly optimizes token prediction (what-to-unmask), inference-time unmasking orders (where-to-unmask) are typically determined by heuristic confidence measures or trained through reinforcement learning with costly on-policy rollouts. To address this, we introduce Gt-Margin, a position-wise score derived from ground-truth tokens, defined as the probability margin between the correct token and its strongest alternative. Gt-Margin yields an oracle unmasking order that prioritizes easier positions first under each partially masked state. We demonstrate that leveraging this oracle unmasking order significantly enhances final generation quality, particularly on logical reasoning benchmarks. Building on this insight, we train a supervised unmasking planner via learning-to-rank to imitate the oracle ordering from masked contexts. The resulting planner integrates into standard MDLM sampling to select where-to-unmask, improving reasoning accuracy without modifying the token prediction model.

Hikaru Asano, Ryo Yonetani, Taiki Sekii et al.

This paper presents Text2Traj2Text, a novel learning-by-synthesis framework for captioning possible contexts behind shopper's trajectory data in retail stores. Our work will impact various retail applications that need better customer understanding, such as targeted advertising and inventory management. The key idea is leveraging large language models to synthesize a diverse and realistic collection of contextual captions as well as the corresponding movement trajectories on a store map. Despite learned from fully synthesized data, the captioning model can generalize well to trajectories/captions created by real human subjects. Our systematic evaluation confirmed the effectiveness of the proposed framework over competitive approaches in terms of ROUGE and BERT Score metrics.

Hikaru Asano, Tadashi Kozuno, Yukino Baba

Recent advances in large language models (LLMs) have yielded impressive performance on various tasks, yet they often depend on high-quality feedback that can be costly. Self-refinement methods attempt to leverage LLMs' internal evaluation mechanisms with minimal human supervision; however, these approaches frequently suffer from inherent biases and overconfidence, especially in domains where the models lack sufficient internal knowledge, resulting in performance degradation. As an initial step toward enhancing self-refinement for broader applications, we introduce an iterative refinement pipeline that employs the Unlabeled-Unlabeled learning framework to improve LLM-generated pseudo-labels for classification tasks. By exploiting two unlabeled datasets with differing positive class ratios, our approach iteratively denoises and refines the initial pseudo-labels, thereby mitigating the adverse effects of internal biases with minimal human supervision. Evaluations on diverse datasets, including low-resource language corpora, patent classifications, and protein structure categorizations, demonstrate that our method consistently outperforms both initial LLM's classification performance and the self-refinement approaches by cutting-edge models (e.g., GPT-4o and DeepSeek-R1).

Hikaru Asano

Bayesian experimental design (BED) has been used as a method for conducting efficient experiments based on Bayesian inference. The existing methods, however, mostly focus on maximizing the expected information gain (EIG); the cost of experiments and sample efficiency are often not taken into account. In order to address this issue and enhance practical applicability of BED, we provide a new approach Sequential Experimental Design via Reinforcement Learning to construct BED in a sequential manner by applying reinforcement learning in this paper. Here, reinforcement learning is a branch of machine learning in which an agent learns a policy to maximize its reward by interacting with the environment. The characteristics of interacting with the environment are similar to the sequential experiment, and reinforcement learning is indeed a method that excels at sequential decision making. By proposing a new real-world-oriented experimental environment, our approach aims to maximize the EIG while keeping the cost of experiments and sample efficiency in mind simultaneously. We conduct numerical experiments for three different examples. It is confirmed that our method outperforms the existing methods in various indices such as the EIG and sampling efficiency, indicating that our proposed method and experimental environment can make a significant contribution to application of BED to the real world.