MNIST : 28*28 grayscale image multi-class classification

In this tutorial we show how green_tsetlin TM can be used to train on the MNIST dataset. MNIST is a benchmark by digit recognition that contains images of handwritten digits with a total of 70,000 images. Each image is a 28x28 pixel grayscale image with values between 0 and 255.

from sklearn.datasets import fetch_openml
from sklearn.model_selection import train_test_split as split

X, y = fetch_openml("mnist_784", version=1, return_X_y=True, as_frame=False)
X_train, X_test, y_train, y_test = split(X, y, test_size=0.2, random_state=42, shuffle=True)

\[(70000, 784) \leftarrow (70000, 28, 28)\]

With sklearn we import an easy to use version of MNIST. This version gives 2d right away hence no flatten is needed. Next, as the TM requires binary values, each pixel is converted with a threshold of 75.

X_train = np.where(X_train > 75, 1, 0)
X_train = X_train.astype(np.uint8)

X_test = np.where(X_test > 75, 1, 0)
X_test = X_test.astype(np.uint8)

y_train = y_train.astype(np.uint32)
y_test = y_test.astype(np.uint32)

We can now train the Tsetlin Machine. Here, it is preferable and recommended to run a hyperparameter search. Head to the IMDB or IRIS tutorial to see how. Parameters set here is from a previous search.

import green_tsetlin as gt

tm = gt.TsetlinMachine(n_literals=28*28,
                       n_clauses=6154,
                       n_classes=10,
                       s=21.627727185060525,
                       threshold=1218)

trainer = gt.Trainer(tm, seed=42, n_epochs=10, n_jobs=1, k_folds=2)

trainer.set_train_data(X_train, y_train)
trainer.set_eval_data(X_test, y_test)

trainer.train()

results = trainer.results

print(results)

{
'best_eval_score': 0.9918857142857143,
'k_folds': 2,
'train_time_of_epochs':
[41.99848390498664,
30.74028508097399,
30.064865624008235,
...
27.486269582004752,
27.345113909977954,
26.679121892957482],
'train_log':
[0.8924285714285715,
0.9485714285714286,
0.9587142857142857,
...
0.9949428571428571,
0.9950571428571429,
0.9953714285714286],
'eval_log':
[0.9393714285714285,
0.9510857142857143,
0.9575428571428571,
...
0.9827714285714285,
0.9822285714285715,
0.9806285714285714]
}