These are my assignments for Practical Machine Deel Learning Course CS4930 at AUC. I try to classify CIFAR-10 Dataset that contains over 60k images with 10 different classes. The techniques used are:
1- KNN and Linear Least Square:
KNN- ACCR of the testing is: 32.4300%
LLS- ACCR of the testing is: 36.3700%
2- Multi-layer Fully Connected Neural Network (NN) classifier using my own implmentation and Keras API.
Own implementation- ACCR of the testing is: 61.25%
Keras API- ACCR of the testing is: 68.50%
3-Convolutional Neural Network (CNN) and Res-Net classifier using my own implmentation and Keras API.
Own implementation- ACCR of the testing is: 82.86%
Keras API- ACCR of the testing is: 92.73%
Note: Some of the codes were adapted from Stanford CS231n Course's tempelates.