Yoshitaka Ushiku
Research Subjects
Research Interests
Yoshitaka Ushiku has a total of 90 co-authors. Here are the top co-authors by number of shared publications:
- Atsushi Hashimoto
OMRON SINIC X, Tokyo, Japan
13 shared publications - Naoya Chiba
Graduate School of Information Sciences, Tohoku University, Sendai, Japan
6 shared publications - Hirotaka Kameko
Kyoto University, Kyoto-shi, Kyoto, Japan
5 shared publications - Kanta Ono
Department of Applied Physics, The University of Osaka, Osaka, Japan
5 shared publications - Shigeo Yoshida
Omron Sinic X Corporation, Tokyo, Japan
3 shared publications Shinsuke Mori
3 shared publications
Latest Publications
- Open Access
- 27 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-25
Cooperative Design Optimization through Natural Language Interaction
- 27 September 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-25
Open AccessCooperative Design Optimization through Natural Language Interaction
- 22 August 2025
- Published by Springer Nature
- p. 90-101
CRNLI: A Textual Entailment Dataset in the Chemistry Domain
- 22 August 2025
- Published by Springer Nature
- p. 90-101
CRNLI: A Textual Entailment Dataset in the Chemistry Domain
- Open Access
- 17 June 2025
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 6 (3)
Bridging Text and Crystal Structures: Literature-driven Contrastive Learning for Materials Science
Understanding structure–property relationships is an essential yet challenging aspect of materials discovery and development. To facilitate this process, recent studies in materials informatics have sought latent embedding spaces of crystal structures to capture their similarities based on properties and functionalities. However, abstract feature-based embedding spaces are human-unfriendly and prevent intuitive and efficient exploration of the vast materials space. Here we introduce Contrastive Language--Structure Pre-training (CLaSP), a learning paradigm for constructing crossmodal embedding spaces between crystal structures and texts. CLaSP aims to achieve material embeddings that 1) capture property- and functionality-related similarities between crystal structures and 2) allow intuitive retrieval of materials via user-provided description texts as queries. To compensate for the lack of sufficient datasets linking crystal structures with textual descriptions, CLaSP leverages a dataset of over 400,000 published crystal structures and corresponding publication records, including paper titles and abstracts, for training. We demonstrate the effectiveness of CLaSP through text-based crystal structure screening and embedding space visualization.
- 17 June 2025
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 6 (3)
Open AccessBridging Text and Crystal Structures: Literature-driven Contrastive Learning for Materials Science
Understanding structure–property relationships is an essential yet challenging aspect of materials discovery and development. To facilitate this process, recent studies in materials informatics have sought latent embedding spaces of crystal structures to capture their similarities based on properties and functionalities. However, abstract feature-based embedding spaces are human-unfriendly and prevent intuitive and efficient exploration of the vast materials space. Here we introduce Contrastive Language--Structure Pre-training (CLaSP), a learning paradigm for constructing crossmodal embedding spaces between crystal structures and texts. CLaSP aims to achieve material embeddings that 1) capture property- and functionality-related similarities between crystal structures and 2) allow intuitive retrieval of materials via user-provided description texts as queries. To compensate for the lack of sufficient datasets linking crystal structures with textual descriptions, CLaSP leverages a dataset of over 400,000 published crystal structures and corresponding publication records, including paper titles and abstracts, for training. We demonstrate the effectiveness of CLaSP through text-based crystal structure screening and embedding space visualization.
- Top Cited PapersOpen Access
- 27 May 2025
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 4 (6) , 1384-1403
Self-driving laboratories in Japan
This perspective highlights the current state of self-driving laboratories (SDLs) in Japan. It also outlines how national funding, community collaboration, and industry support their growth, along with future directions.
- 27 May 2025
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 4 (6) , 1384-1403
Top Cited PapersOpen AccessSelf-driving laboratories in Japan
This perspective highlights the current state of self-driving laboratories (SDLs) in Japan. It also outlines how national funding, community collaboration, and industry support their growth, along with future directions.
- Open Access
- 25 April 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
VisDev: A Stroke Deviation Visualization System Using Point-Set Registration for Handwriting Training
- 25 April 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessVisDev: A Stroke Deviation Visualization System Using Point-Set Registration for Handwriting Training
- Open Access
SBS Figures: Pre-training Figure QA from Stage-by-Stage Synthesized Images
Building a large-scale figure QA dataset requires a considerable amount of work, from gathering and selecting figures to extracting attributes like text, numbers, and colors, and generating QAs. Although recent developments in LLMs have led to efforts to synthesize figures, most of these focus primarily on QA generation. Additionally, creating figures directly using LLMs often encounters issues such as code errors, similar-looking figures, and repetitive content in figures. To address this issue, we present SBSFigures (Stage-by-Stage Synthetic Figures), a dataset for pre-training figure QA. Our proposed pipeline enables the creation of chart figures with complete annotations of the visualized data and dense QA annotations without any manual annotation process. Our stage-by-stage pipeline makes it possible to create diverse topic and appearance figures efficiently while minimizing code errors. Our SBSFigures demonstrate a strong pre-training effect, making it possible to achieve efficient training with a limited amount of real-world chart data starting from our pre-trained weights.
Open AccessSBS Figures: Pre-training Figure QA from Stage-by-Stage Synthesized Images
Building a large-scale figure QA dataset requires a considerable amount of work, from gathering and selecting figures to extracting attributes like text, numbers, and colors, and generating QAs. Although recent developments in LLMs have led to efforts to synthesize figures, most of these focus primarily on QA generation. Additionally, creating figures directly using LLMs often encounters issues such as code errors, similar-looking figures, and repetitive content in figures. To address this issue, we present SBSFigures (Stage-by-Stage Synthetic Figures), a dataset for pre-training figure QA. Our proposed pipeline enables the creation of chart figures with complete annotations of the visualized data and dense QA annotations without any manual annotation process. Our stage-by-stage pipeline makes it possible to create diverse topic and appearance figures efficiently while minimizing code errors. Our SBSFigures demonstrate a strong pre-training effect, making it possible to achieve efficient training with a limited amount of real-world chart data starting from our pre-trained weights.
- 8 December 2024
- Published by Springer Nature
- p. 175-193
Learning 3D Point Cloud Registration as a Single Optimization Problem
- 8 December 2024
- Published by Springer Nature
- p. 175-193
Learning 3D Point Cloud Registration as a Single Optimization Problem
- 21 November 2024
- Published by Springer Nature
- p. 123-140
COM Kitchens: An Unedited Overhead-View Video Dataset as a Vision-Language Benchmark
- 21 November 2024
- Published by Springer Nature
- p. 123-140
COM Kitchens: An Unedited Overhead-View Video Dataset as a Vision-Language Benchmark
- Open Access
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 9234-9243
AdaCoder: Adaptive Prompt Compression for Programmatic Visual Question Answering
Visual question answering aims to provide responses to questions given visual input. Recently, visual programmatic models (VPMs), which generate programs to answer questions through large language models (LLMs), have attracted attention. However, they often require long input prompts to provide the LLM with sufficient API usage details to generate relevant code. To address this limitation, we propose AdaCoder, an adaptive prompt compression framework for VPMs. AdaCoder operates in two phases: a compression phase and an inference phase. In the compression phase, given a preprompt that describes all API definitions with example code snippets, a set of compressed preprompts is generated, each depending on a specific question type. In the inference phase, AdaCoder predicts the question type and chooses the appropriate corresponding compressed preprompt to generate code to answer the question. In experiments, we apply AdaCoder to ViperGPT and demonstrate that it reduces token length by 71.1%, while maintaining or even improving the performance of visual question answering.
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 9234-9243
Open AccessAdaCoder: Adaptive Prompt Compression for Programmatic Visual Question Answering
Visual question answering aims to provide responses to questions given visual input. Recently, visual programmatic models (VPMs), which generate programs to answer questions through large language models (LLMs), have attracted attention. However, they often require long input prompts to provide the LLM with sufficient API usage details to generate relevant code. To address this limitation, we propose AdaCoder, an adaptive prompt compression framework for VPMs. AdaCoder operates in two phases: a compression phase and an inference phase. In the compression phase, given a preprompt that describes all API definitions with example code snippets, a set of compressed preprompts is generated, each depending on a specific question type. In the inference phase, AdaCoder predicts the question type and chooses the appropriate corresponding compressed preprompt to generate code to answer the question. In experiments, we apply AdaCoder to ViperGPT and demonstrate that it reduces token length by 71.1%, while maintaining or even improving the performance of visual question answering.
- Open Access
- 16 September 2024
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 3 (10) , 2130-2136
Force-controlled robotic mechanochemical synthesis
This study introduces a robotic mechanochemical synthesis system with controlled force. It enhances reproducibility and controls reaction pathways by adjusting grinding force and speed, opening new possibilities in mechanochemical synthesis.
- 16 September 2024
- Published by Royal Society of Chemistry (RSC) in Digital Discovery
- Vol. 3 (10) , 2130-2136
Open AccessForce-controlled robotic mechanochemical synthesis
This study introduces a robotic mechanochemical synthesis system with controlled force. It enhances reproducibility and controls reaction pathways by adjusting grinding force and speed, opening new possibilities in mechanochemical synthesis.