liveBook · Manning

Handwritten digit classification

As explained in Deep Learning with R, Third Edition

Handwritten digit classification is the task of assigning an image of a handwritten digit to its corresponding numeric category.

Handwritten digit classification assigns each 28 × 28 grayscale image to one of ten numeric classes, from 0 through 9. MNIST provides 60,000 training images and 10,000 test images, making the task a practical baseline for checking whether a neural-network workflow works. section 2.1

A working classification pipeline follows these steps:

  • Load MNIST through Keras, keeping images and their corresponding labels in training and test arrays. section 2.1

Reading a prediction

The model returns ten probabilities for each test image. Selecting the largest probability gives the predicted class; one example assigns class 8 a probability of 0.9993394, while the recorded label is 7. Evaluation across the test set reports 0.9784 accuracy.

How Deep Learning with R, Third Edition covers it

How deep it goes
Demonstrates the complete data-to-evaluation workflow, but does not develop the underlying steps beyond later explanation.
Examples
Uses MNIST images, Keras arrays, a dense network, probability outputs, and a held-out accuracy result.
What you can do after
Builds and evaluates a basic handwritten-digit classifier with Keras and interprets its class-probability predictions.
How it connects
Connects image classification with neural networks, Keras, MNIST, labels, training sets, test sets, and softmax probabilities.
Cover of Deep Learning with R, Third Edition
liveBookmanning.com
Read Deep Learning with R, Third Edition on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Deep Learning with R, Third Edition. The section links point to the parts of the book it draws on.