Group work on MultiLayer Perceptron (MLP) and Hyperparameters Optimization.
See IS-ML-Take-Home-Assignment-2.pdf for tasks and instructions. See Group_13-ML_Assignment_2.pdf for our report.
Part A includes:
- Implementation of a MultiLayer Perceptron (MLP) model from scratch, which can perform a two-class classification task on the given dataset.
- Activation Functions: Rectified Linear Unit (ReLU) and Sigmoid activation function
- Binary Cross-Entropy Loss function
- Mean Square error loss
- Huber loss
Part B includes:
- Confusion Matrix
- Training and Validation Losses
- Preprocessing Techniques: Normalization and One-Hot Encoding
- Hyperparameter Tuning
- Heatmap Visualization