Weekly summary 2016-08-13

Goals:

• Run basic PMI experiments
  • PMI images, PMI idea
  • Get dictionary learning/weighted SVD code from Yingyu.
  • Write up findings.
• Learning
  • Go through half of ATAP.
  • Ch. 4 of HDP.
  • Understand conjugate gradients using Chebyshev polys. Look at AGD from this point of view.
  • Summarize bandit paper.
• Representation learning
  • Clarify idea.
• Tensorflow
  • Get basic LSTM running.
  • Play with parameters in MNIST.

Summary:

• Got PMI code running.
• Approximation theory and approximation practice: Notes
  • Understood conjugate gradient bound using Chebyshev polynomials (@Cyril). Notes
• Talk: Yuanzhi on Bandit convex optimization (FOLLOW UP)
• Representation learning summary. Talked to Yingyu and kernel DL.
• Read BH16.