Lei Sha

Lei Sha

Research Associate

Leaving date: 31st March 2022

Themes:

• Artificial Intelligence and Machine Learning

See Also:

• Main page

Interests

Controllable information extraction, Controllable text generation, Natural language understanding

Biography

I am a Postdoctoral researcher advised by Prof. Thomas Lukasiewicz in the Intelligent System Lab, University of Oxford. Previously, I was an NLP research scientist in Apple. While at Apple, I am responsible for Siri’s module, such as domain classification and chit-chat dialogue. Before that, in 2018, I obtained my PhD degree and graduated from Peking University, China, where I worked with Prof. Zhifang Sui, Prof. Baobao Chang, and Prof. Sujian Li. I also served as a research assistant in Microsoft Research Asia, working with Chin-yew Lin, Lintao Zhang, and Qi Chen. During my PhD period, I majored in information extraction and text generation. I have published many first-author papers in top conferences of NLP, such as ACL, EMNLP, NAACL, AAAI, etc. My research interest focus on natural language understanding and controllable text generation.

Selected Publications

View AllManage publications

• A Stable‚ Fast‚ and Fully Automatic Learning Algorithm for Predictive Coding Networks

  Tommaso Salvatori‚ Yuhang Song‚ Yordan Yordanov‚ Beren Millidge‚ Lei Sha‚ Cornelius Emde‚ Zhenghua Xu‚ Rafal Bogacz and Thomas Lukasiewicz

  In Proceedings of the 12th International Conference on Learning Representations‚ ICLR 2024‚ Vienna‚ Austria‚ 7–11 May 2024. May, 2024.

  Details about A Stable‚ Fast‚ and Fully Automatic Learning Algorithm for Predictive Coding Networks | BibTeX data for A Stable‚ Fast‚ and Fully Automatic Learning Algorithm for Predictive Coding Networks

• Rationalizing Predictions by Adversarial Information Calibration

  Lei Sha‚ Oana−Maria Camburu and Thomas Lukasiewicz

  In Artificial Intelligence. Vol. 315. Pages 103828. February, 2023.

  Details about Rationalizing Predictions by Adversarial Information Calibration | BibTeX data for Rationalizing Predictions by Adversarial Information Calibration | Link to Rationalizing Predictions by Adversarial Information Calibration

• Text attribute control via closed−loop disentanglement

  Lei Sha and Thomas Lukasiewicz

  In Transactions of the Association for Computational Linguistics (TACL). 2023.

  Accepted for publication

  Details about Text attribute control via closed−loop disentanglement | BibTeX data for Text attribute control via closed−loop disentanglement | Link to Text attribute control via closed−loop disentanglement

Supervisor