What question did this study set out to answer?

The aim is to apply machine learning techniques to enhance digit recognition accuracy using the MNIST dataset.

June 15, 2026Open Access

Pattern Recognition of Handwritten Digits Mnist Dataset

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial compares the accuracy of MLP and CNN in recognizing handwritten digits, suggesting improvements in training methods.

The aim is to apply machine learning techniques to enhance digit recognition accuracy using the MNIST dataset.
Utilized two neural network architectures: a three-layer MLP with ReLU and dropout, and a stack of CNN with ReLU and max pooling.
Analyzed the training speed and accuracy of each approach during testing with MNIST data.
Implemented modifications to the CNN to achieve higher accuracy rates.
MLP achieved 95% accuracy during training and 94% during testing.
CNN achieved 99% accuracy during training and 98.9% during testing; modifications improved this to 99.82% training and 99.25% testing accuracy.

Meshaal Mouawad (2021) studied this question.

View Full Paper

Ask AI

Bookmark