Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Aug 1, 2018

• Merge pull request #1 from sachag678/final-run

  Show description for 9037636
  sachag678

  authored
  9037636

  Copy full SHA for 9037636

• Merge branch 'master' into final-run

  sachag678

  authored
  6ef1178

  Copy full SHA for 6ef1178

• Finshed the implementation using keras backend, and custom loss function for the policy gradient method using a NN. The current parameters work well. Further tuning of hyperparameters could be useful.

  sachag678

  committed
  7a1057b

  Copy full SHA for 7a1057b

Commits on Jul 28, 2018

• First attempt at implementing cartpole using policy gradient.

  sachag678

  committed
  af089c4

  Copy full SHA for af089c4

Commits on Jul 10, 2018

• Updated to handle non-square matrices.

  sachag678

  committed
  bf68f09

  Copy full SHA for bf68f09

• put main code into main.

  sachag678

  committed
  412198f

  Copy full SHA for 412198f

• updated comments

  sachag678

  committed
  1897f01

  Copy full SHA for 1897f01

• ex google search problem - find the longest run of colours adjacent to each other in a matrix.

  sachag678

  committed
  c36821e

  Copy full SHA for c36821e

Commits on Jul 1, 2018

• Test to simulate binary classification problem using naive bayes to show how class imbalance affects the posterior probability.

  sachag678

  committed
  e0e2a3c

  Copy full SHA for e0e2a3c

Commits on Jun 12, 2018

• Updated gitignore

  sachag678

  committed
  2f0706d

  Copy full SHA for 2f0706d

• Updated gitignore

  sachag678

  committed
  1be9f91

  Copy full SHA for 1be9f91

• Changes to tic-tac-toe

  sachag678

  committed
  7f31d4c

  Copy full SHA for 7f31d4c

• Simple NN

  sachag678

  committed
  5b2af5f

  Copy full SHA for 5b2af5f

• insertion sort example.

  sachag678

  committed
  b25e50b

  Copy full SHA for b25e50b

• Ai Assitant

  sachag678

  committed
  d805f4c

  Copy full SHA for d805f4c

Commits on Apr 24, 2018

• bagging vs wagging.

  sachag678

  committed
  f560538

  Copy full SHA for f560538

• Parralel and model averaging.

  sachag678

  committed
  82327b4

  Copy full SHA for 82327b4

Commits on Apr 10, 2018

• simple multithread example using factorial.

  sachag678

  committed
  6fa7703

  Copy full SHA for 6fa7703

Commits on Dec 17, 2017

• ran 100000 iterations, and no improvement. Not sure what the issue is

  sachag678

  committed
  04d8288

  Copy full SHA for 04d8288

• Added two different brains for each player. Still doesn't work. Maybe not enough training? Or does it have something to do with the 5 and 2 we pass to the network.

  sachag678

  committed
  ce5879e

  Copy full SHA for ce5879e

Commits on Dec 16, 2017

• Used a neural network to approximate the state-action function and learned how to play against itself. Proceeded to draw

  sachag678

  committed
  aa6fefd

  Copy full SHA for aa6fefd

• Plays the game against itself. However due to the fact that we don't consider any states the play is not very optimal

  sachag678

  committed
  e7f770d

  Copy full SHA for e7f770d

• Added slightly better expert who has a winning strategy if the opponent plays v. poorly.

  sachag678

  committed
  b00b763

  Copy full SHA for b00b763

Commits on Dec 13, 2017

• Update readme.md

  sachag678

  authored
  d3c69de

  Copy full SHA for d3c69de

Commits on Dec 12, 2017

• Changed the method keeping track of the visited states from a list to a dict to avoid duplicates

  sachag678

  committed
  b67fdec

  Copy full SHA for b67fdec

• Changed the way the rewards are updated to be at the end of the epoch. Also set it up so each instance of visiting a state was given a reward

  sachag678

  committed
  a6fbf75

  Copy full SHA for a6fbf75

• Added notes about game and type of learning

  sachag678

  committed
  7323d39

  Copy full SHA for 7323d39

• Performed ploicy evaluation using model free Monte Carlo Every Visit updates and plotted the optimal value function

  sachag678

  committed
  82f54d0

  Copy full SHA for 82f54d0

Commits on Dec 6, 2017

• Created a gitignore

  sachag678

  committed
  9f5ef0c

  Copy full SHA for 9f5ef0c

• added commens and docstrings

  sachag678

  committed
  f6f9b88

  Copy full SHA for f6f9b88

• Used the value iteration with the bellman optimality equation. Solved the control problem

  sachag678

  committed
  a354eb4

  Copy full SHA for a354eb4

• Corrected the Bellman Expecation equation

  sachag678

  committed
  e8c9363

  Copy full SHA for e8c9363

• removed old version

  sachag678

  committed
  18a3d4c

  Copy full SHA for 18a3d4c

• Changed name

  sachag678

  committed
  ab2082d

  Copy full SHA for ab2082d

• changed name

  sachag678

  committed
  d6a450e

  Copy full SHA for d6a450e