SCBI: A GPU-accelerated initialization strategy that replaces random weights with statistical estimates. "Solves" linear layers before training starts, reducing initial loss by 90% on regression tasks.

A machine learning model to to predict whether it will rain tomorrow in Australia or not.

This project analyzed and compared the performance of 16 machine learning models on a supervised classification task using the Dry Bean dataset. This project pursued two objectives: (1) Measure how accurately each model classifies unseen bean samples, and (2) Determine each model’s runtime for its classification training and testing process.