Cypaubr / ML-k-NN-Digit-Recognition

Hand written digits recognition using machine learning

Machine Learning Digit Recognition

Recognizes digits using the k-NN algorithm and the MNIST datasets for training and testing.

Steps

1. The MNIST dataset is loaded
2. The 70000 pieces dataset is samplet into parts of 5000
3. The dataset is splitted into training & testing sets
4. Error is tested using a given k-NN (3 in the code)
5. Score is optimized by testing various k values and holding the best score
6. The prediction is given by the model using the k value of the best score

Example

Score optimization

For this example the best score is 3 and the worse score is 13.

Prediction using the best score

Prediction using the worse score