Senior Capstone

An Analysis of Hyperparameter Fine-tuning on Small-Scale Face Image Recognition with PyTorch

The goal of this research project is to provide an overview on the development of a small-scale face recognition system with the PyTorch machine learning library, as well as the effects of hyperparameter fine-tuning on model performance. A total of 27 convolutional neural network models will be created across three different optimizers (Adam, SGD, RMSProp), learning rates (0.001, 0.0005, 0.0001), and dropout rates (0.1, 0.3, 0.5). With a fixed number of convolutional layers and training epochs, the loss and accuracy for training and validation data will be collected and analyzed.