Hands-On Machine Learning with TensorFlow.js

Sequential model

With the sequential model API, you can construct the model by piling up each layer. Here, we create the multilayer perceptron or deep learning model, which can be described as the graph of each neural layer. For example, a model that recognizes an input vector with 784 elements to predict the output of 10 categories can be represented as follows:

We can use the following code to build a sequential model that connects the two fully connected layers that are shown in the preceding diagram. The sequential model can be expressed as a list of each layer:

const model = tf.sequential({ layers: [ tf.layers.dense({inputShape: [784], ...