Bruno Castro da Silva

Professor da Silva is interested in designing reinforcement learning (RL) algorithms that can solve large sets of diverse real-life problems, while ensuring that the learning process is safe according to criteria defined by a designer. To achieve this goal, his research focuses primarily on two key problems: (1) how to design general-purpose RL algorithms capable of autonomously decomposing complex tasks into simpler sub-problems, for which specialized reusable and composable skills be can be learned; and (2) how to ensure that these skills are learned in a way that meets user-specified safety requirements with high probability. These are fundamental questions that underlie the gap between what artificial intelligence agents can - in principle - do and what we can effectively get them to do given our current algorithms. 

The ultimate goal of Professor da Silva's work is to design the necessary tools so that reinforcement learning algorithms can be widely used to solve challenging real-world tasks in homes and in the workplace, in a safe way, and with as little human intervention as possible. 

More broadly, Professor da Silva's research interests lie in the intersection of machine learning, reinforcement learning, optimal control theory, and robotics, and include the construction of hierarchical policies, active learning, open-ended learning, biologically-plausible intrinsic motivation mechanisms, Bayesian optimization applied to control, and machine learning algorithms with high-probability safety and fairness guarantees.