Xuefeng WEI

Publications

CArtBench: A Museum-Grounded Benchmark for Evidence-Faithful Chinese Art Understanding

Xuefeng Wei, Wang Zhixuan, Xuan Zhou, Zhi Qu, Hongyao Li, Yusuke Sakai, Hidetaka Kamigaito, Taro Watanabe

Published in Findings of EMNLP, 2026 (to appear)

This paper introduces a museum-grounded benchmark for evaluating evidence-faithful understanding and interpretation of Chinese artworks. Read more

Where Do Vision-Language Models Fail on Abstract Reasoning? An Information Processing Analysis of ARC

Xuan Zhou, Xuefeng Wei, Zhi Qu, Yusuke Sakai, Hidetaka Kamigaito, Taro Watanabe

Published in EMNLP Main Conference, 2026 (to appear)

This paper analyzes the information-processing failures of vision-language models on abstract visual reasoning tasks. Read more

ARTKGC: Leakage-Controlled Multimodal Knowledge Graph Completion for Artwork Explanation

Xuan Zhou, Xuefeng Wei, Yuta Kato, Shintaro Ozaki, Kazuki Hayashi, Yusuke Sakai, Hidetaka Kamigaito, Katsuhiko Hayashi, Taro Watanabe

Published in Findings of EMNLP, 2026 (to appear)

This paper presents a leakage-controlled multimodal knowledge graph completion framework for artwork explanation. Read more

“Yuki Gets Sushi, David Gets Steak?”: Uncovering Gender and Racial Biases in LLM-Based Meal Recommendations

Xuefeng Wei, Xuan Zhou, Yusuke Sakai, Taro Watanabe

Published in EACL Main Conference, 2026

This paper uncovers gender and racial biases in meal recommendations produced by large language models. Read more

A Cross-Domain Tool-Augmented Vision-Language Framework for Remote Sensing Image Understanding

Xuan Zhou, Xuefeng Wei, Zhi Qu, Yusuke Sakai, Hidetaka Kamigaito, Taro Watanabe

Published in Remote Sensing, 2026

This paper develops a tool-augmented vision-language framework for robust cross-domain remote sensing image understanding. Read more

FLDNet: A Foreground-Aware Network for Polyp Segmentation Leveraging Long-Distance Dependencies

Xuefeng Wei, Xuan Zhou

Published in ICONIP, 2023

This paper proposes a foreground-aware network that leverages long-distance dependencies for polyp segmentation. Read more