MNIST database : http://yann.lecun.com/exdb/mnist/
- training stage
- validation stage
- test stage
Done getting input header.
Successfully read and stored 60000 feature vectors.
Done getting input label header.
Successfully read and stored 60000 feature labels.
Training data size: 4500
Test data size: 1200
Validation data size: 300
Successfully Extracted 10 Unique Classes.
Starting with Kth-nearest algorithms
check kth-nearest, k = 1
Validation performance for K = 1 : 94%
check kth-nearest, k = 2
Validation performance for K = 2 : 94.6667%
check kth-nearest, k = 3
Validation performance for K = 3 : 96.3333%
check kth-nearest, k = 4
Validation performance for K = 4 : 95.6667%
Testing performance = 94.0833%