Intel Deep Learning Community of Practice talk

Video Description

Social learning helps humans and animals rapidly adapt to new circumstances, coordinate with others, and drives the emergence of complex learned behaviors. What if it could do the same for AI? This talk focuses on Social Reinforcement Learning, which leverages social learning and affective computing to enhance coordination, learning, generalization, and human-AI interaction. To improve coordination, we give agents an intrinsic motivation to increase their causal influence over the actions of other agents, and show that this leads to the emergence of communication and enhances cooperation. Beyond coordination, I will demonstrate how multi-agent training can be a useful tool for improving learning and generalization. I will present PAIRED, in which an adversary learns to construct training environments to maximize regret between a pair of learners, leading to the generation of a complex curriculum of environments that improve the learner’s zero-shot transfer to unknown, single-agent test tasks. I will present techniques which enable agents to learn socially from other agents co-existing in their environment, including humans. However, learning from humans requires recognizing their social and affective cues. Therefore I will discuss how to dramatically enhance the accuracy of affect detection models using personalized multi-task learning to account for inter-individual variability. Together, this work argues that Social RL is a valuable approach for developing more general, sophisticated, and cooperative AI, which is ultimately better able to serve human needs.