Pinned Loading

1. random_forest_cluster random_forest_cluster Public

   A clustering algorithm, where a forest of shallow trees are trained on random subsets of the features. Points are clustered together according to how often they end up on the same leaf.

   Python 8 2

2. regret_descent regret_descent Public

   A decoration on any gradient-based optimization algorithm. Optimization parameters (e.g. learning rate) are updated after each step toward what we might have wished them to be for the last step. Pr…

   Python

3. impact_measure impact_measure Public

   Preliminary inquiry in impact measures for RL agents within MDPs. See Amodei, et al., (2016) for an introducton to the topic.

4. predict_slow_algorithm predict_slow_algorithm Public

   Takes a series that is being extended slowly, and approximates the value to be expected a few iterations in the future. Could potentially be used to speed up an optimization procedure, for example.

   Python

5. truth_table_neural_net truth_table_neural_net Public

   This never succeeded in outperforming a basic neural net, and took significantly longer to train. A neural network with the capacity for "truth table layers" is implemented here.

   Python