2. Learning in neural systems

We saw how neurons can do meaningful operations in the last module. But, given that neural networks potentially consist of millions of neurons, how do we configure them to do meaningful work? And, what if circumstances change? Can they adapt?

In this module, we will look at learning theory and understand how we can train neurons towards certain objectives.

Lecture content

1. Learning in neural networks (slideshow)

2. Supervised learning

3. Project 2: Predicting digits from audio

Literature

• Training Spiking Neural Networks Using Lessons From Deep Learning

• SuperSpike: Supervised Learning in Multilayer Spiking Neural Networks

• Neuromorphic Engineering: In Memory of Misha Mahowald