Much research in machine learning has focused on designing algorithms for discovering knowledge from data and learning a target model. Machine teaching leverages this work to optimally teach machines and humans. For example, consider a situation where a “teacher” already knows the learning goal and wants to steer the “learner” towards this target as quickly as possible. What is the optimal set of demonstrations and lessons to provide? Similar ideas are used in a very different setting, namely, adversarial attacks on machine learning systems, where an attacker or hacking algorithm (the “teacher”) manipulates a machine learning system (the “learner”) by maliciously modifying the training data. This lecture will provide an overview of machine teaching and cover the following three aspects: (i) how machine teaching differs from machine learning, (ii) the problem space of machine teaching, and (iii) recent work on developing teaching algorithms for human learners.
About the lecturer
Dr. Adish Singla
Adish Singla is a tenure-track faculty at the Max Planck Institute for Software Systems (MPI-SWS), where he leads the Machine Teaching research group. He received a Ph.D. from ETH Zurich in 2017. Before starting his Ph.D., he worked as a Senior Development Lead in Bing Search for over three years. He is a recipient of the Facebook fellowship in the area of machine learning, Microsoft Research Tech-Transfer award, and Microsoft Gold Star award.