About me

I'm Keqian Li, and I am a Researcher at Meta AI. I obtained my Ph.D. at the School of Engineering, University of California Santa Barbara under the Regent Scholarship, advised by Prof. Xifeng Yan (Chair), Prof. William Wang and Prof. Ambuj Singh, and have been affiliated with Google, Facebook, MSR, Yahoo! Labs and AOL in the past. Before that, I finished my Bachelor of Engineering the Special Pilot CS Class (Yao Class) supervised by Andrew Yao and Bachelor of Economics from Tsinghua University School at Economic and Management at Tsinghua University.

I'm broadly interested in foundations and applications of representation learning. Specifically, I have been working on the software-hardware codesign for composing large capacity AI models to push the boundary of approximation, optimization and generalization to improve the efficiency and scalability of Meta's monetization products.

What's New

  • 02/2023: We're hiring AI Research interns to work broadly in the area of pretraining with possible long term extensions. Interns intersted in both industry and academdia are welcome to apply.
  • 07/2022: Exicited to join Meta AI, see you back in California!
  • 03/2022: Exicited to file 3 patents on large scale application of AutoML, shout out to legal team for the quick correspondance along the way.

Publications

  • CALM: Commen-Sense Knowledge Augmentation for Document Image Understanding
    Qinyi Du, Qinging Wang, Keqian Li, Jidong Tian, Liqiang Xiao, Yaohui Jin
    Proceedings of the 30th ACM International Conference on Multimedia, 3282-3290.     [paper]
  • SuperCone: Modeling Heterogeneous Experts with Concept Meta-learning for Unified Predictive Segments System
    Keqian Li, Yifan Hu
    CoRR abs/2203.07029..     [paper][patent documents]
  • MGEL: Multigrained Representation Analysis and Ensemble Learning for Text Moderation
    Fei Tan, Changwei Hu, Yifan Hu, Kevin Yen, Z Wei, A Pappu, S Park, Keqian Li
    IEEE Transactions on Neural Networks and Learning Systems.     [paper][patent documents]
  • Hadoop-MTA: a system for multi data-center trillion concepts Auto-ML atop Hadoop
    Keqian Li, Yifan Hu, Manisha Verma, Fei Tan, Changwei Hu, Tejaswi Kasturi, Kevin Yen
    2021 IEEE International Conference on Big Data (Big Data), 5953-5955.     [paper]
  • BAN: Large Scale Brand ANonymization for Creative Recommendation via Label Light Adaptation
    Keqian Li, Kevin Yen, Shaunak Mishra, Yifan Hu, Manisha Verma
    2021 IEEE International Conference on Big Data (Big Data), 5953-5955.     [paper]
    (Winner of Techpulse 2021 Best Internal Talk Award)
  • TNT: Text normalization based pre-training of transformers for content moderation
    Fei Tan, Yifan Hu, Changwei Hu, Keqian Li, Kevin Yen
    Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP).     [paper][data and code]
  • Hiercon: Hierarchical organization of technical documents based on concepts
    Keqian Li, Shiyang Li, Semih Yavuz, Hanwen Zha, Yu Su, Xifeng Yan
    2019 IEEE International Conference on Data Mining (ICDM), 379-388.     [paper][data and code]
    (2019 IEEE International Conference on Data Mining Best Paper candidate)
  • Mining algorithm roadmap in scientific publications
    Hanwen Zha, Wenhu Chen, Keqian Li, Xifeng Yan
    Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.     [paper][data and code] [website]
  • Mining and analyzing technical knowledge based on concepts
    Keqian Li
    Phd. Thesis. University of California, Santa Barbara..     [paper]
  • Concept mining via embedding
    Keqian Li, Hanwen Zha, Yu Su, Xifeng Yan
    2018 IEEE International Conference on Data Mining (ICDM), 267-276.     [paper][data and code]
  • Unsupervised neural categorization for scientific publications
    Keqian Li, Hanwen Zha, Yu Su, Xifeng Yan
    Proceedings of the 2018 SIAM International Conference on Data Mining, 37-45.     [paper][data and code]
  • FTS: Faceted Taxonomy Construction and Search for Scientific Publications
    Hanwen Zha, Jiaming Shen, Keqian Li, W Greiff, Michelle Vanni, Jiawei Han, Xifeng Yan
    Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.     [paper][data and code] [website]
  • PoQaa: Text Mining and Knowledge Sharing for Scientific Publications
    Keqian Li, Ping Zhang, Honglei Liu, Hanwen Zha, Xifeng Yan. Proc. of Int. Conf. on Knowledge Discovery and Data Mining (KDD 2018). (demo) [paper][video]
  • Discovering enterprise concepts using spreadsheet tables
    Keqian Li, Yeye He, Kris Ganjam
    Proceedings of the 23th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.     [paper][patent documentation]
  • On social event organization
    Keqian Li, Wei Lu, Smriti Bhagat, Laks V.S. Lakshmanan, Cong Yu
    Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.     [paper]
    (Ranked #1 for research publication for event organization by Google Scholar)

Patents

  • System and method for text moderation via pretrained transformers
    US Patent Granted US11,481,543
  • Determining a hierarchical concept tree using a large corpus of table values
    US Patent Granted US10,789,229
  • System And Method For Integrated Large Scale Audience Targeting Via Augmented Heterogeneous Sub Systems
    US Patent Filed US827,364
  • System And Method For Augmenting Existing Experts For Enhanced Predictions
    US Patent Filed US827,400
  • System And Method For Integrating Multiple Expert Predictions In A Nonlinear Framework Via Learning
    US Patent Filed US827,431

Services

  • Associate Editor/Reviewer : IEEE Transactions on Knowledge and Data Engineering, IEEE Transactions on Computational Social Systems , Journal of Shanghai Jiao Tong University, Conference on Neural Information Processing, Conference on Neural Information Processing Systems, ACM Multimedia

Contact Me

Email:      keqianli [at] meta.com

Address:  1 Hacker Way, Menlo Park, CA, 94025