About me

Keep Learning & Be Positive!

Lijun Wu is currently a Senior Researcher of Machine Learning Group in Microsoft Research Asia (MSRA). He got the Ph.D. degree from Sun Yat-sen University (SYSU), School of Data and Computer Science in 2020, was a member of joint Ph.D. program between SYSU and MSRA, advised by Dr. Tie-Yan Liu and Prof. Jianhuang Lai. He received MSRA Ph.D. Fellowship in 2018. Prior to that, he obtained the Bachelor degree in the same department in 2015.

His researches focus on Deep Learning, NLP, Multimodality Learning, Medical Health, Bio-NLP, Bio-Embedding, Drug Discovery. Works during the Ph.D. journey are mostly on sequence learning tasks such as neural machine translation. He is also interested in meta learning, reinforcement learning.

You can also refer to the Microsoft Page.

Feel free and welcome to contact for intern positions and possible collobaration!

Selected Research

  • Consistency training and dropout
    • R-Drop (sub-model consistency)
    • UniDrop (unified dropout)
    • JANUS (NAT&AT consistency)
    • R^2-DDI (drug-drug interaction consistency)
    • C^2-Rec (recommendation consistency)
  • Drug discovery and bioinformatics
    • SPRoBERTa (local fragment-based protein pre-training)
    • R^2-DDI (drug-drug interaction consistency)
    • DMCG (direct molecular conformation generation)
    • SMT-DTA (semi-supervised drug-target affinity prediction)
    • AbBERT-HMPN (pre-training for antibody design)
  • Neural Machine Translation
    • RL4NMT (the first RL for NMT survey)
    • NAT beyond (a comprehensive study of NAT and beyond)
    • BERT-NMT (BERT for NMT)
    • ANMT (adversarial NMT)
    • SCA (soft contextual data augmentation)
    • Mono-NMT (large scale monolingual data for NMT)


11/25/2022 Our R$^2$-DDI about Drug-Drug Interaction (DDI) is accepted by Briefings in Bioinformatics.
11/19/2022 Our AMOM about NAT training and one paper about retrosynthesis prediction are accepted by AAAI-2023.
10/15/2022 Our DMCG about molecule conformation generation is accepted by TMLR.
10/07/2022 Our JANUS about NAT&AT training and one paper about temporal sequence generation are accepted by EMNLP-2022.
09/01/2022 Our SPRoBERTa about protein pre-training is accepted by Briefings in Binformactics.
06/18/2022 One paper about Unified 2D and 3D molecule pretraining and one paper about retrosythetic prediction are accepted by KDD-2022.
05/20/2022 We summarize a comprehensive survey on Non-Autoregressive Generation, check here!