Artificial Neural Network from Scratch (MNIST)
Other
Built and trained an artificial neural network from scratch using the MNIST dataset. The work involved preparing training inputs and evaluating model performance across multiple customizable neural network architectures. It focused on implementing the training loop, learning rate, number of hidden-layer neurons, and iterations rather than manual label annotation. • Implemented multi-architecture ANN training and testing using MNIST. • Tuned hyperparameters such as hidden layer size and learning rate. • Performed evaluation by running test passes and comparing results. • Validated a no-framework approach (no PyTorch/TensorFlow).
2020 - 2023