SVM-Softmax-Classifier-CIFAR10-dataset

Required libraries

• Numpy
• scipy
• Keras
• matplotlib

Running the program

python runSvmSoftmax.py

SVM

Best validation accuracy: 39.7
Best Model parameter, lr = 6e-07, reg = 18000
Training acc: 37.5938%
Validating acc: 39.7%
Testing acc: 36.76%

Softmax

Best validation accuracy: 38.1
Best Model parameter, lr = 8e-07, reg = 12000
Training acc: 37.86938775510204%
Validating acc: 38.1%
Testing acc: 36.68%