Fashion-MNIST Classification & Hyperparameter Optimization

• Implemented a full neural network from scratch using NumPy, including forward and backward

propagation.

• Built and evaluated a Keras-based deep learning model achieving ~89% test accuracy.

• Conducted comprehensive model evaluation using accuracy metrics, confusion matrices, and

classification reports.

• Applied Random Search and Genetic Algorithms to optimize hyperparameters and improve

validation performance.

• Analyzed training dynamics and optimization trade-offs, demonstrating practical ML

experimentation skills.