Next week we have a guest from the United States.
Dr. Yuxin Chen, an assistant professor in the Department of Computer Science, University of Chicago, will talk about Tractability, Interpretability, and Robustness in Machine Learning.
Abstract: Machine teaching studies the interaction between a (machine) teacher and a learner, where the teacher’s objective is to find an optimal training sequence to steer the learner towards a target concept. It provides a rigorous formalism for a number of real-world applications including personalized educational systems, adversarial attacks, and imitation learning. In this talk, I will discuss the algorithmic challenges in modeling and teaching realistic (typically human) learners. I will start by introducing the machine teaching problem under the well studied “version space” learner's model, and show that finding the optimal set of training examples amounts to a combinatorial optimization problem which is NP-hard. I will then focus on more realistic learner’s models, and discuss our recent work in teaching certain classes of limited capacity learners (e.g., learners with limited memory and/or limited computational/perceptual capability).
Short bio of the speaker:
Yuxin Chen is an assistant professor of computer science at the University of Chicago, where he leads the interactive learning systems group. Previously, he was a postdoctoral scholar in the Department of Computing and Mathematical Sciences at the California Institute of Technology, and received his Ph.D. degree in computer science from ETH Zurich. His research interest covers broadly the design, analysis, and implementation of novel machine learning algorithms for probabilistic reasoning, interactive machine learning, and decision making. More specifically, he works on resource-efficient, robust, and interpretable learning systems that actively extract information, identify the most relevant data and make effective decisions under uncertainty.
You can join the seminar via the following link:
19 October, 2020, 15:00
19 October, 2020, 16:00