-
This is my final project submission for my Machine Learning class.
-
The task was to use two different Machine Learning techniques to fill in the 300 by 300 blacked-out portion in the center of the following images:
-
I implemented a neural network class and decision tree class from scratch, including gradient descent, backpropagation, correlation calculation, and highest expected information calculation algorithms.
-
Then, I trained an ensemble of Neural Networks to iteratively predict each pixel in the masked image and leveraged Bootstrapping by using past outputs to make future outputs.
-
I employed a similar technique for the decision tree model.
For full details and the results, check out the project!