Supervised Approaches to CIFAR10

Supervised machine learning is a widely used form of artificial intelligence. There are plenty of ways to approach supervised learning: Some of them being Neural Networks, Convolutional Neural Networks and Residual Networks. In this repository we develop an in depth analysis of the difference between these on the CIFAR10 dataset using Jupyter Notebooks and Pytorch.

Deep Linear NN

Testing Loss: 1.516066481353371 Testing Accuracy: 0.4617999792098999

Deep CNN

Testing Loss: 1.1094991602715414 Testing Accuracy: 0.6746999621391296

Deep CNN - Data Augmentation

Testing Loss: 0.8757611672589733 Testing Accuracy: 0.7076999545097351

Pre-trained ResNet18 - 32x32 pixel images

Testing Loss: 0.7395002558163017 Testing Accuracy: 0.8202999830245972

Pre-trained ResNet18 - Resized 224x224 pixel images

Testing Loss: 0.18219238568014304 Testing Accuracy: 0.9521999955177307