- Reached an accuracy of 88% for MNIST by just linear classification.
- Uses matrix multiplication of linear regression to predict whether an image is cat or a dog
- Vizualize the weights of the equation Y = mX + c, i.e. m, to see how the model is classifying
- The main purpose was to learn how tensor addition and multiplication can be used to solve hardest of problems.
- If you are interested in learning more about image classification, I have implemented the same problem using CNN and Transfer learning. https://github.com/bipinkc19/ImageClassifier-Cat-Dog-
- Linear regression from scratch using ordinary least error and gradient descent.