Biography
I am a Ph.D. student at the Nara Institute of Science and Technology (NAIST), where I delve into the intricacies of Natural Language Processing (NLP) at the NLP Lab. My primary focus is on representation learning, especially for Knowledge Graph and Language Modeling. I am also enthusiastic about fusing NLP techniques with other modalities, e.g., audio generation.
E-mail: feng.xincan.fy2 [at] is.naist.jp
Links: Google Sites, LinkedIn, Twitter, ZhiHu, GitHub, ORCID, Google Scholar, Semantic Scholar
Publications
Proceedings
Xincan Feng, Hidetaka Kamigaito, Katsuhiko Hayashi and Taro Watanabe. "Unified Interpretation of Smoothing Methods for Negative Sampling Loss Functions in Knowledge Graph Embedding", workshop RepL4NLP held in conjunction with ACL 2024. [paper] [code] [poster]
Xincan Feng and Akifumi Yoshimoto. "Llama-VITS: Enhancing TTS Synthesis with Semantic Awareness", LREC-COLING 2024. [paper] [code] [poster] [slides]
Xincan Feng, Hidetaka Kamigaito, Katsuhiko Hayashi and Taro Watanabe. "Model-based Subsampling for Knowledge Graph Completion", IJCNLP-AACL 2023. [paper] [code] [poster]
Xincan Feng, Zhi Qu, Yuchang Cheng, Taro Watanabe and Nobuhiro Yugami. "Sharing Parameter by Conjugation for Knowledge Graph Embeddings in Complex Space", workshop TextGraphs-16 held in conjunction with COLING 2022. [paper] [code] [slides] [video]
Not Refereed
Xincan Feng. "Model-based Subsampling for Knowledge Graph Completion", Journal of Natural Language Processing, 2024, Volume 31, Issue 1, Pages 287-293. [article]
林 和樹, 出口 祥之, Xincan Feng, 上垣外 英剛, 林 克彦, 渡辺 太郎. "kNN-LMによる知識グラフを用いた大規模言語モデルにおける知識の操作", NLP若手の会第18回シンポジウム (YANS 2023). (第一著者奨励賞)
Xincan Feng, 上垣外 英剛, 林 克彦, 渡辺 太郎. "知識グラフ補完のためのモデル予測に基づくサブサンプリング", 言語処理学会第29回年次大会 (NLP 2023). [paper]
Professional Activities
Invited Speeches
2023.03 Invited by the Data Science Group of Korea Advanced Institute of Science & Technology (KAIST), I got my first chance of being an invited speaker. In this one-hour talk, I presented our newest research on model-based subsampling.
Workshops
2023.12 I participated in the NAIST Touch Stone Workshop 2023, which was part of the JST SPRING program. During this workshop, our team had the opportunity to present our innovative business idea, which was supervised and received feedback from both professors and a delegation of business leaders.
Program Committee
2024 TextGraphs-17, Reviewer
2023 TNNLS, Reviewer
Education
2023.04 - Present Ph.D. student in Information Science and Engineering, Nara Institute of Science and Technology
Affiliated with Natural Language Processing Laboratory (NLP Lab).
Supervised by Prof. Taro Watanabe and A.P. Hidetaka Kamigaito.
2021.04 - 2023.03 M.Eng. in Information Science and Engineering, Nara Institute of Science and Technology
Affiliated with Natural Language Processing Laboratory (NLP Lab) and Multilingual Knowledge Computing Laboratory (MKC Lab).
Supervised by Prof. Taro Watanabe, Prof. Nobuhiro Yugami, A.P. Yuchang Cheng, and A.P. Hidetaka Kamigaito.
2011.09 - 2015.07 B.S. in Natural Science, Minzu University of China
Majored in Information and Computing Science, School of Science
Work Experiences
2023.08 - Present AI Research Scientist, Audio Team, AI Lab, CyberAgent, Inc.
Research on text-to-speech AI.
2023.04 - 2024.03 Research Assistant, Nara Institute of Science and Technology.
Research on knowledge graph.
2023.04 - 2023.06 R&D Associate, R&D-LLM Team, AKARI Inc.
Automating dataset construction of construction laws and regulations in Japan.
Developing language model application system using LangChain.
2021.12 - 2022.01 Research Intern, Artificial Intelligence Laboratory, Fujitsu.
Surveying and implementing variants for BioBERT and checking the impact on the performance.
Related paper: An Nguyen Le, Hajime Morita and Tomoya Iwakura. Learning Entity-Likeness with Multiple Approximate Matches for Biomedical NER, 2021. [paper]
2021.08 - 2022.01 Mathematical Formula Annotator, Miyao Group, The University of Tokyo.
Preparing and annotating mathematical formulas in the paper data, and analyzing and checking bugs in the annotation tool.
Related paper: Takuto Asakura, Yusuke Miyao, Akiko Aizawa, Michael Kohlhase. MioGatto: A Math Identifier-oriented Grounding Annotation Tool. In 13th MathUI Workshop at 14th Conference on Intelligent Computer Mathematics (MathUI 2021). [preprint] [code]
Scholarships・Grants
2023.04 - 2026.03 奈良先端科学技術大学院大学「先端科学技術融合分野におけるイノベーティブ博士人材支援プロジェクト (NAIST Touch Stone) 」.
JST SPRING Grant Number JPMJSP2140. / JST次世代研究者挑戦的研究プログラム JPMJSP2140.
2013 Academy Prize of College of Science, Minzu University of China.
Skills
Programming Languages: PyTorch, Python
Tools: Linux, Git, LaTeX, MATLAB, GeoGebra, Office
Languages: Chinese (native), English (business), Japanese (fluent)