Wei Zhang  Wei Zhang photo       

contact information

Research Engineer, AI Foundations Lab



Wei Zhang is a research engineer at IBM Research AI. His current focus is Artificial Neural Networks and deep learning, especially neural networks with external memory and/or attention for reasoning and inference. He is interested in following topics:

DL for NLP: machine reading comprehension, knowledge graph based question answering with deep neural networks;     

life-long-learning, transfer and multi-task learning;    

learning to learn, model based optimization, curriculum learning, active learning;   

algorithm learning, program induction, text and logic reasoning;   

variational and bayesian methods combined with Deep learning.

Before IBM Research, he worked as a team lead at IBM Watson on using deep learning and machine learning for improving Watson services. He worked on machine reading comprehension, and his submission was one of the earliest on the Stanford Question Answering Dataset leaderboard. He also participated in internal Watson competitions on factoid question answering, and as a team won the first place.

In 2015, he became interested in Neural networks with memory, and starts to explore such kind of models since then. In 2014, he works on using large scale textual data for applications such as language modeling and spell correction.

From 2012 to 2014, he works as a research assistant at CMU LTI where he received his masters of AI in 2014, with Dr. Judith Gelernter on machine learning based spatial and temporal information extraction and disambiguation. He is the creater of CMU GeoLocator, an opensource, location geoparsing and geocoding tool.

Before 2012, he spent some wonderful years in Beijing working on Information retrieval, Ontology construction, Natural language processing. He received his Bachelor of CS from Harbin Institute of Technology in 2007 where he started his interest in Natural language processing.

Wei can also be found here but not here anymore.

Technical Areas