Center for Advanced Intelligence Project

21 papers have been accepted at the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), a top conference for Computational Linguistics (July 2- 7, 2026, San Diego, California). For more details, please refer to the link below.

[ACL 2026] https://2026.aclweb.org/

[Main Conference]

• The Confidence Dichotomy: Analyzing and Mitigating Miscalibration in Tool-Use Agents
  Weihao Xuan (University of Tokyo / RIKEN AIP)
  Qingcheng Zeng (Northwestern University)
  Heli Qi (RIKEN AIP / Waseda University)
  Yunze Xiao (Carnegie Mellon University)
  Junjue Wang (University of Tokyo)
  Naoto Yokoya (University of Tokyo / RIKEN AIP)
• PHOTON: Hierarchical Autoregressive Modeling for Lightspeed and Memory-Efficient Language Generation
  Yuma Ichikawa (Fujitsu Limited / RIKEN AIP)
  Naoya Takagi (Fujitsu Limited)
  Takumi Nakagawa (Fujitsu Limited / Institute of Science Tokyo)
  Yuzi Kanazawa (Fujitsu Limited)
  Akira Sakai (Fujitsu Limited /Tokai University)
• AgentCoMa: A Compositional Benchmark Mixing Commonsense and Mathematical Reasoning in Real-World Scenarios
  Lisa Alazraki(Imperial College London)
  Lihu Chen(Imperial College London)
  Ana Brassard(RIKEN AIP)
  Joe Stacey(University of Sheffield)
  Hossein A. Rahmani(University College London)
  Marek Rei(Imperial College London).
• Cell-Based Representation of Relational Binding in Language Models
  Qin Dai(Tohoku University)
  Benjamin Heinzerling(RIKEN AIP / Tohoku University)
  Kentaro Inui(MBZUAI / Tohoku University / RIKEN AIP)
• Corpus-Dependent Subcharacter Encoding via HMM-Guided Code Assignment
  Tatsuya Hiraoka (MBZUAI / NAIST / RIKEN AIP)
• Fine-Grained Analysis of Shared Syntactic Mechanisms in Language Models
  Ryoma Kumon (University of Tokyo / RIKEN AIP)
  Hitomi Yanaka (University of Tokyo / RIKEN AIP)
• How to Improve LLMs’ Performance on Specific Languages: A Perspective on LLMDerived Language Similarity
  Xinhe Shi (Zhejiang University)
  Qingcheng Zeng (Northwestern University)
  Weihao Xuan (University of Tokyo / RIKEN AIP)
  Linchao Zhu (Zhejiang University)
• Memorization, Emergence, and Explaining Reversal Failures: A Controlled Study of Relational Semantics in LLMs
  Yihua Zhu  (Kyoto University / NII)
  Qianying Liu  (Kyoto University / NII)
  Jiaxin Wang  (Kyoto University)
  Fei Cheng (Kyoto University) Chaoran Liu (NII)
  Akiko Aizawa (University of Tokyo / NII)
  Sadao Kurohashi (Kyoto University / NII)
  Hidetoshi Shimodaira (Kyoto University / RIKEN AIP)
• The Hidden Costs and Measurement Gaps of Reinforcement Learning with Verifiable Rewards
  Fang Wu (Stanford University)
  Aaron Tu (University of California, Berkeley)
  Weihao Xuan (University of Tokyo / RIKEN AIP)
  Heli Qi (RIKEN AIP / Waseda University)
  Xu Huang (Georgia Institute of Technology)
  Qingcheng Zeng (Northwestern University)
  Shayan Talaei (Stanford University)
  Yijia Xiao (University of California, Los Angeles)
  Peng Xia (The University of North Carolina at Chapel Hill)
  Xiangru Tang (Yale University)
  Yuchen Zhuang (Georgia Institute of Technology)
  Bing Hu (Independent)
  Hanqun Cao (Chinese University of Hong Kong)
  Wenqi Shi (University of Texas Southwestern Medical Center)
  Rui Yang (National University of Singapore)
  Nan Liu (National University of Singapore)
  Huaxiu Yao (The University of North Carolina at Chapel Hill)
  Ge Liu (University of Illinois at Urbana-Champaign)
  Li Erran Li (Columbia University)
  Amin Saberi (Stanford University)
  Naoto Yokoya (University of Tokyo / RIKEN AIP)
  Jure Leskovec (Stanford University)
  Yejin Choi (Stanford University / NVIDIA)
• When Benchmarks Leak: Inference-Time Decontamination for LLMs
  Jianzhe Chai (Science Tokyo)
  YU ZHE (RIKEN AIP)
  Jun Sakuma (RIKEN AIP / Science Tokyo)
• Why Mean Pooling Works: Quantifying Second-Order Collapse in Text Embeddings
  Tomomasa Hara (Tohoku University)
  Hiroto Kurita (Tohoku University)
  Masaaki Imaizumi (University of Tokyo / RIKEN AIP)
  Kentaro Inui (MBZUAI / Tohoku University / RIKEN AIP)
  Sho Yokoi (NINJAL / Tohoku University / RIKEN AIP))

[Findings]

• Applicability Condition Extraction for Therapeutic Drug-Disease Relations
  Guanting Luo (RIKEN AIP / Osaka University)
  Noriki Nishida (RIKEN AIP)
  Yuji Matsumoto (RIKEN AIP / Tohoku University)
  Yuki Arase (Institute of Science Tokyo / RIKEN AIP)
• Code-Switching Information Retrieval: Benchmarks, Analysis, and the Limits of Current Retrievers
  Qingcheng Zeng (Northwestern University)
  Yuheng Lu (Waseda University)
  Zeqi Zhou (Brown University)
  Heli Qi (RIKEN AIP / Waseda University)
  Puxuan Yu (Snowflake)
  Fuheng Zhao (Snowflake)
  Hitomi Yanaka (University of Tokyo / RIKEN AIP)
  Weihao Xuan (University of Tokyo / RIKEN AIP)
  Naoto Yokoya (University of Tokyo / RIKEN AIP)
• Differentially Private Synthetic Text Generation for Retrieval-Augmented Generation (RAG)
  Junki Mori (NEC)
  Kazuya Kakizaki (NEC)
  Taiki Miyagawa (NEC)
  Jun Sakuma (RIKEN AIP / Science Tokyo)
• Establishing a Scale for Kullback-Leibler Divergence in Language Models Across Various Settings
  Ryo Kishino  (Kyoto University)
  Yusuke Takase (NII)
  Momose Oyama  (Kyoto University / RIKEN AIP)
  Hiroaki Yamagiwa  (Kyoto University)
  Hidetoshi Shimodaira (Kyoto University / RIKEN AIP)
• Pattern Enhanced Multi-Turn Jailbreaking: Exploiting Structural Vulnerabilities in Large Language Models
  Ragib Amin Nihal (RIKEN AIP / Science Tokyo)
  Rui Wen (Science Tokyo)
  Kazuhiro Nakadai (Science Tokyo)
  Jun Sakuma (RIKEN AIP / Science Tokyo)
• Sentipolis: Emotion-Aware Agents for Social Simulations
  Chiyuan Fu (Carnegie Mellon University)
  Lyuhao Chen (Carnegie Mellon University)
  Yunze Xiao (Carnegie Mellon University)
  Weihao Xuan (University of Tokyo / RIKEN AIP)
  Carlos Busso (Carnegie Mellon University)
  Mona T. Diab (Carnegie Mellon University)
• Timesteps of Mamba Align with Human Reading Times
  Yuji Yamamoto (SOKENDAI / NINJAL))
  Shinnosuke Isono (NINJAL)
  Yoshinobu Kawahara (University of Osaka / RIKEN AIP)
  Sho Yokoi (NINJAL / Tohoku University / RIKEN AIP)
• ToolGrad: Efficient Tool-use Dataset Generation with Textual “Gradients”
  Zhongyi Zhou (University of Tokyo / RIKEN AIP / Google)
  Kohei Uehara (University of Tokyo)
  Haoyu Zhang (University of Tokyo)
  Jingtao Zhou (Google)
  Lin Gu (RIKEN AIP / University of Tokyo)
  Ruofei Du (Google)
  Zheng Xu (Google)
  Tatsuya Harada (University of Tokyo / RIKEN AIP)
• VLURes: Benchmarking Long-Text Grounding and Cross-Lingual Robustness in Vision Language Models
  Jesse Atuhurra (NAIST)
  Iqra Ali (QMUL)
  Tomoya Iwakura (Meiji University)
  Hidetaka Kamigaito (NAIST)
  Tatsuya Hiraoka (MBZUAI / NAIST / RIKEN AIP)
• What Do Vision–Language Models Encode for Personalized Image Aesthetics Assessment?
  Koki Ryu (University of Tokyo / RIKEN AIP)
  Hitomi Yanaka (University of Tokyo / RIKEN AIP)

Related Laboratories