- Python
- numpy
- Pandas
- MNIST Dataset
- sklearn
We use the MNIST handwritten letters dataset and use KNN to calculate the accuracy of prediction. Current accuracy : 90%
To run the code : python knn.py
Modify the values of k inside to change accuracy levels.
Contributions are welcome!
We use MNIST training and test data to calculate an optimal value for b to minimize the cost function.
To run the code : python lr.py
Modify the value for learningRate to try other possible answers.
We use MNIST dataset and do 10-Fold Cross Validation to get training and test data. We find optimal weights and calculate the value of TPR and FPR to plot the ROC curve.
To run the code : python logistic-regression.py
Modify the value for learning_rate, gradient_steps and splits to tinker with it.
We used MNIST dataset to predict the hand written digit. The neural network has one hidden Softmax layer, the activation function of input layer is Sigmoid() and the the loss function is Mean Squared Error.
- To run the code :*
python main.py
Modify the value for learning_rate, epochs, batch_size and gradient_steps to tinker with it.