Applying MLP to the MNIST dataset

Let's apply MLP to the MNIST dataset.

We'll cover the following

Code example

Let us discuss our first deep neural network for handwritten character recognition, using the MNIST\text{MNIST} dataset. This dataset has dominated much of the early DNN\text{DNN} developments and must be considered a classic by now. The data set consists of digitized examples of handwritten numbers from 0 to 9. Each image consists of 28×28 28 \times 28 pixels. An example is shown as input in the figure below. This benchmark dataset has 60,00060,000 training examples and 10,00010,000 test examples. A developer is free to choose any portion of the 60,00060,000 training images as the validation set.

Get hands-on with 1200+ tech skills courses.