BCS541 Neuroscience-inspired Artificial Intelligence (2023 Fall)
[Important notice] This course is the second part of the Brain x AI lecture series:
How AI and the brain work (undergraduate course BCS441 (previously BiS 429):
focuses on inverse problems. It covers linear methods, deep learning, and neuroscience of deep learning.Neuroscience-inspired artificial intelligence (graduate course BCS541):
focuses on control problems. It covers biological error backpropagation, temporal credit assignment problems, reinforcement learning theory/algorithms, neuroscience of reinforcement learning, and brain-inspired reinforcement learning.I will no longer teach BiS 429 and BCE 772.
Summary: Error backpropagation through time in AI and brains
The course aims to understand theory, algorithms, and neuroscience of error backpropagation through time. The lecture covers error backpropagation in biological neural networks, temporal credit assignment, reinforcement learning algorithms, and neuroscience of reinforcement learning. The last section includes striatal model-free and cortical model-based learning, prefrontal meta-reinforcement learning, entorhinal goal-driven learning, hippocampal successor representation, and dopaminergic distributional value coding. Students are expected to master necessary skills to make the best use of machine learning theory to understand how and why the brain learns.
