View on GitHub

yunhe.github.io

Yun He

Yun He(贺赟)

  • I am a Senior Research Scientist at Meta (Facebook).
  • Email: heyunyun2014 'at' gmail.com
  • Research Interests: Natural Language Processing, Recommender Systems and Information Retrieval. Multi-task Learning. Multi-objective Optimization.



Howdy! I am a senior research scientist at Meta AI. I was a member of the infolab led by Prof. James Caverlee in the Computer Science & Engineering Department at Texas A&M University. I'm broadly interested in the area of natural language processing, information retrieval and recommender systems.

Previously, I was at the ICA lab in the School of Computer Science and Software Engineering at East China Normal University and worked with Dr. Qinmin Hu and Dr. Liang He. I obtained my Bachelor's degree from Donghua University.

Research Experience

Senior Research Scientist, Meta AI. (Aug 2023 – Present)

Research Scientist, Meta AI. (July 2022 – July 2023)

Research Project: Multi-objectives and multi-task optimization balancing

Research Intern, Google Brain. (September 2021 – December 2021)

Mentor: Dr. Huaixiu Zheng and Dr. Yi Tay
Research Project: A novel prompt-tuning method for transformer-based multi-task co-training

Research Intern, Facebook AI. (May 2021 – August 2021)

Mentor: Dr. Xinyi Zhang
Research Project: User history modeling for recommender system.

Research Intern, Facebook AI. (June 2020 – September 2020)

Mentor: Dr. Xue Feng
Research Project: Automatic balancing multiple training loss terms to improve multi-task learning-based recommender system.

Research Assistant, Texas A&M University (Fall 2017 – present)

Advisor: Dr. James Caverlee
Research Interests: Natural Language Processing and Recommender System

  • Research Project 3 (New): Infusing disease knowledge into BERT for Health Question Answering, Medical Inference and Disease Name Recognition (will appear in EMNLP’2020)

  • Research Project 2 (New): We propose a new dataset for Paraphrase Identification Requiring Computer Science Domain Knowledge (will appear in EMNLP’2020)

  • Research Project 1: Recommending user-generated item lists (e.g. playlists on Spotify and book lists on Goodreads) (CIKM’2019) and Automatic Continuation (next item prediction) for user-generated item lists (WSDM’2020)

  • Research experience in East China Normal University (2014 – 2017)

    Advisor: Dr. Qinmin Hu and Dr. Liang He
    Research Interests: Information Retrieval in Biomedical Domain (ECIR’2016)
    1st place in Clinical Decision Support Track Task B, Text Retrieval Conference (TREC’2015)
    1st place in CLEF eHealth Task2, Conference and Labs of the Evaluation Forum (CLEF’2015)

    Publications

    See also my Google Scholar profile.

    • PromptAttack: Probing Dialogue State Trackers with Adversarial Prompts [pdf]
      (ACL) (findings), 2023
      Xiangjue Dong, Yun He, Ziwei Zhu, James Caverlee

    • HyperPrompt: Prompt-based Task-Conditioning of Transformers [pdf]
      International Conference on Machine Learning (ICML), 2022
      Yun He, Huaixiu Steven Zheng, Yi Tay, Jai Gupta, Yu Du, Vamsi Aribandi, Zhe Zhao, YaGuang Li, Zhao Chen, Donald Metzler, Heng-Tze Cheng and Ed H. Chi

    • MetaBalance: Improving Multi-Task Recommendations via Adapting Gradient Magnitudes of Auxiliary Tasks [pdf] [code]
      The WebConference (WWW), 2022
      Yun He, Xue Feng, Cheng Cheng, Geng Ji, Yunsong Guo, James Caverlee

    • Popularity Bias in Dynamic Recommendation
      Knowledge Discovery and Data Mining (KDD), 2021
      Ziwei Zhu, Yun He, Xing Zhao and James Caverlee

    • Vibe Check: Social Resonance Learning for Enhanced Recommendation.
      The IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM 2021)
      Yin Zhang, Yun He and James Caverlee

    • Item Relationship Graph Neural Networks for E-Commerce
      IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2021
      Weiwen Liu, Yin Zhang, Jianling Wang, Yun He, James Caverlee, Patrick P. K. Chan, Daniel S. Yeung and Pheng-Ann Heng

    • Popularity-Opportunity Bias in Collaborative Filtering
      WSDM 2021
      Ziwei Zhu, Yun He, Xing Zhao, Yin Zhang, Jianling Wang, and James Caverlee

    • Infusing Disease Knowledge into BERT for Health Question Answering, Medical Inference and Disease Name Recognition [pdf] [code]
      EMNLP 2020
      Yun He, Ziwei Zhu, Yin Zhang, Qin Chen and James Caverlee

    • PARADE: A New Dataset for Paraphrase Identification Requiring Computer Science Domain Knowledge [pdf] [code]
      EMNLP 2020
      Yun He, Zhuoer Wang, Yin Zhang, Ruihong Huang and James Caverlee

    • Consistency-Aware Recommendation for User-Generated Item Lists Continuation. [pdf] [code]
      International Conference on Web Search and Data Mining (WSDM 2020)
      Yun He, Yin Zhang, Weiwen Liu and James Caverlee

    • Unbiased Implicit Recommendation and Propensity Estimation via Combinational Joint Learning (short)
      Recsys 2020
      Ziwei Zhu, Yun He, Yin Zhang, and James Caverlee

    • Content-Collaborative Disentanglement Representation Learning for Enhanced Recommendation
      Recsys 2020
      Yin Zhang, Ziwei Zhu, Yun He, and James Caverlee

    • Adaptive Hierarchical Translation-based Sequential Recommendation.
      The Web Conference (WWW 2020)
      Yin Zhang, Yun He, Jianling Wang, James Caverlee.

    • A Hierarchical Self-Attentive Model for Recommending User-Generated Item Lists. [pdf] [code]
      International Conference on Information and Knowledge Management (CIKM 2019).
      Yun He, Jianling Wang, Wei Niu and James Caverlee

    • Self-Attention based Network For Query Expansion in Medical Domain.
      International Joint Conference on Neural Networks (IJCNN 2019).
      Su Chen, Qinmin Hu, Yang Song, Yun He, Huaying Wu and Liang He

    • Pseudo-Implicit Feedback for Alleviating Data Sparsity in Top-K Recommendation. (acceptance rate: 20%) [pdf] [code]
      IEEE International Conference on Data Mining (ICDM 2018).
      Yun He, Haochen Chen, Ziwei Zhu and James Caverlee.

    • Estimating Probability Density of Content Types for Promoting Medical Records Search. (acceptance rate: 21%)
      European Conference on Information Retrieval (ECIR 2016).
      Yun He, Qinmin Hu, Yang Song and Liang He.

    • ECNU at 2016 eHealth Task 1: Handover Information Extraction.(Working Notes) [pdf]
      Conference and Labs of the Evaluation Forum (CLEF 2016).
      Yang Song, Yun He, Hongyu Liu, Qinmin Hu, Liang He, Yueyao Wang.

    • ECNU at 2016 eHealth Task 3: Patient-centred Information Retrieval. (Working Notes) [pdf]
      Conference and Labs of the Evaluation Forum (CLEF 2016).
      Yang Song, Yun He, Hongyu Liu, Yueyao Wang, Qinmin Hu, Liang He and Guihua Luo.

    • ECNU at TREC 2016: Web-based query expansion and experts diagnosis in Medical Information Retrieval. (Working Notes) [pdf]
      Text Retrieval Conference (TREC 2016).
      Hongyu Liu, Yang Song, Yun He, Yueyao Wang, Qinmin Hu, Liang He.

    • ECNU at 2015 CDS track: two re-ranking methods in medical information retrieval. (Working Notes) [pdf]
      Text Retrieval Conference (TREC 2015).
      1st place in Clinical Decision Support Track (Task B)
      Yang Song, Yun He, Qinmin Hu, Liang He.

    • ECNU at 2015 eHealth Task 2: User-centred Health Information Retrieval. (Working Notes) [pdf]
      Conference and Labs of the Evaluation Forum (CLEF 2015).
      1st place in CLEF eHealth (Task2)
      Yang Song, Yun He, Qinmin Hu, Liang He, E Mark Haacke.

    • ECNU at TREC 2014: Clinical decision support track. (Working Notes)
      Text Retrieval Conference (TREC 2014).
      Mingyao Li, Yang Song, Yun He, Qinmin Hu, Liang He, E Mark Haacke.

    Industry Engineering Experience

    • Software engineering intern, Enterprise Data Services (EDS) team, Cisco, Shanghai, China, May 2016 - July 2016.
      Area: Information Retrieval
      Project: Searching for Duplicate Registrations of Customers.

    Teaching

    Technical Talks

    • Pseudo-Implicit Feedback for Alleviating Data Sparsity in Top-K Recommendation. ICDM, Singapore, 2018.
    • Location-Sensitive User Profiling Using Crowdsourced Labels. AAAI, New Orleans, USA, 2018. (On behalf of Wei Niu, James Caverlee and Haokai Lu)
    • Estimating Probability Density of Content Types for Promoting Medical Records Search. ECIR, Italy, 2016.

    Professional Services

    • External Reviewer: WWW'(18), WSDM 2020
    • Reviewer for journals: SNAM2019; TKDE 2018, 2019; Neurocomputing 2021; IEEE Intelligent Systems 2021; Information Fusion 2022; Natural Language Engineering 2022
    • Reviewer for conferences: ACL Demo Track 2020, 2021, 2022; EMNLP 2021, 2022; NAACL demo Track 2022; WSDM 2022; Neurips 2022; AAAI 2023;

    Awards

    • ECIR Student Travel Award, 2016
    • National Scholarship for Graduate Student, 2016
    • National Second Prize in China Graduate Mathematical Modeling Contest, 2014
    • Honorable Mention in International Interdisciplinary Contest in Modeling, 2013
    • National Second Prize in National College Mathematical Contest in Modeling

    Personal

    • Met my wife Jingwen Han in a badminton court at ECNU.
    • A big fan of badminton, football and history.