The core goal of this project is to develop methods that help a computer not only have the ability to continuously learn from data, but also apply human knowledge well in the learning process. Those methods are capable of working with large data or infinite data streams. This project (led by Associate Professor Dr. Than Quang Khoat) is part of a series of collaborative research between Hanoi University of Science and Technology, Kyoto University (Japan) and the University of Oregon (USA).
This project is expected to have a wide impact:
(1) In theory: the project seeks a new direction to solve some fundamental challenges when helping computers continuously learn from data streams. The project investigates new ways to encode human or external knowledge into a model so that machines will learn and reason better from (infinite) data sequences. Ultimately, the project’s research provides effective solutions to some of the core problems of future Artificial Intelligence.
(2) In practice: many applications have to work on continuously incoming data streams. Meanwhile, human knowledge is ubiquitous and provides an excellent source for coping well with many challenges. The research in this project will merge the two in a unified way to increase the performance of machine learning models when applied in many fields, such as online advertising, recommender systems, genetic analysis, diagnosis, …
(3) Regarding training: the project will contribute to training many graduate students, master’s students, and students.
