20BM6JP38_Assign5
Various configurations of CNN are used to classify images from CIFAR-10 dataset
Created on June 1|Last edited on June 1
Comment
Model 1: SGD optimizer, learning rate = 0.001, momentum = 0.9, cross-entropy loss
Model 2: Adam optimizer, learning rate = 0.01, cross-entropy loss
Model 3: SGD optimizer, learning rate = 0.001, momentum = 0.9, squared-error loss
Model 4: Adam optimizer, learning rate = 0.01, squared-error loss
Model 5: Pretrained ResNet34, SGD optimizer, learning rate = 0.01 , momentum = 0.9, cross-entropy loss.
Image Classification using CNN:
Maintain a kernel size of 5 x 5 for the convolutional layers
CNN architecture used for Model 1 to Model 4
Comparison of results:
Run set
5
Confusion matrix across models on test data:
Run set
5
Example test predictions across models:
Run set
5
Add a comment