Summer Lecture List

• Optimization
  • Overview: A brief overview of stochastic gradient descent and then an in depth look at variance reduction optimization and the theoretical and practical convergence rate increase
• Back Propogation Algorithm
  • One of the fundamental algorithm that deep learning is based on. BP is a specific algorithm for computing the gradient of a neural network
• General Adversial Networks
  • Overview: Similar to variational inference, GAN attempt to approximate an intractable probability distribution through a learning process of a game between two adversaries
• Long Short Term Memory Algorithm
  • A famous recurrent neural network algorithm
• Mathematical Introduction to Reinforcement Learning
  • Learning without separation between training and testing. Instant feedback
• Radamacher Complexity
  • Radamacher compexity is a function class measure that gives tighter theoretical bounds on the generalization error of binary margin learners
• Natajaran Dimension
  • An extension of VC dimension to the multiclass classification setting
• Computational Complexity of Learning
  • PAC learner hypothesis classes have a restriction of the big-O runtime of the learning algorithm

References

• Understanding Machine Learning: From theory to algorithms. Shai Shalev-Shwartze and Shai Ben-David