Haim Sompolinsky, Hebrew Univ. Israel and Harvard Univ. USA

Statistical Mechanics of Deep Neural Networks (10.5h)



1. Learning in neural networks: architectures, algorithms, training and generalization

2. Learning in a single layer network: The perceptron and Support Vector Machines

3. Deep Recursive Networks: A minimal model of Deep Neural Networks (DNNs)

4. Classification and geometry of neural object manifolds

5. Object manifolds in artificial and biological DNNs