Thursday, Jun 16

To do

• Complete notes for lectures 12 (Kernels and SVM) and start writing notes for lectures 13
• Start watching Andrew Ng lectures 17 (skip 16)
  • Learning With Large Datasets
  • Stochastic Gradient Descent
  • Mini-Batch Gradient Descent
  • Stochastic Gradient Descent Convergence
  • Online Learning
• Read Karpathy blog post on reinforcement learning in preparation for Adam’s whiteboard lecture on it tomorrow

Done

• Watched Andrew Ng lectures 17 up to and not including the end, which is about map reduce and data parallelism
• Read Karpathy blog post on Policy Gradients and read the Nervana post on Q-Learning
  • All of this is under the field of Reinforcement Learning, which is what we’ll be using to compete in the upcoming VizDoom competition
• Skipped writing notes for lecture 12 since Andrew Ng doesn’t do such a good job in covering them. Instead, I’m using this post on kernels as a learning resource

Extras

• Playing Atari With Deep Reinforcement Learning
• ViZDoom: A Doom-based AI Research Platform for Visual Reinforcement Learning
• Forgot to mention that I bumped into this cool paper a couple of days ago: Convolutional Sketch Inversion. I’d definitely want to read through this soon. Currently held back by note taking the Andrew Ng lectures.