Training progression of the custom CNN on CIFAR-10 data

Image source: https://commons.wikimedia.org/wiki/File:Neural_network_explain.png

Project information

  • Category: Software, Machine Learning
  • Completion date: March 2022
  • Technologies: Machine Learning, CNNs, Python

Details

For ECE C147 at UCLA (Neural Networks and Deep Learning), I built a neural network from scratch using Python. I built support for the following functions so that they could be combined into a custom network

  • Fully connected layers
  • Convolutional layers
  • Max pooling
  • ReLu activation
  • Batch normalization
  • Dropout
  • Softmax

The custom networks were applied to the CIFAR-10 dataset for image recognition tasks