Date & Time: 

   2024 / 3 / 15 (Fri) 14:20 - 16:20

Location: 

   Delta Building R216, NTHU

Speaker: 

   簡仁宗 教授

   國立陽明交通大學 電機工程學系

Topic: 

   Generative Language Models for Conversational AI

Abstract: 

   This talk addresses the fundamentals and advances in construction of generative language models for conversational AI. First, the evolution of language models from n-gram, recurrent neural network, transformer, BERT, GPT and ChatGPT is introduced. As the model size grows exponentially, it becomes crucial to carry out the parameter efficient learning (PEL) in the era of generative pre-trained transformers. Next, a number of practical solutions to PEL for large-scaled GPT in a variety of downstream tasks are presented and surveyed. Finally, this talk describes a recent work on building a mental support dialogue system via spoken language. We will mention how to integrate a system to accomplish a spoken dialogue system with empathy.

Autobiography: 

　Jen-Tzung Chien is the Chair Professor in electrical and computer engineering, and computer science at the National Yang Ming Chiao Tung University, Taiwan. He served as the tutorial speaker in top conferences. Dr. Chien has published extensively, including three books and over 250 peer-reviewed articles, many on machine learning, deep learning, Bayesian learning with applications on natural language processing and computer vision.