Tensorflow.js tf.layers addLoss() 方法

Tensorflow.js 是由谷歌开发的开源库，用于在浏览器或节点环境中运行机器学习模型以及深度学习神经网络。

.addLoss()函数用于将损失附加到所述层。此外，损失可能取决于一些输入张量，例如操作损失取决于所述层的输入。

句法：

addLoss(losses)

参数：

• 损失：这是规定的损失。它可以是RegularizerFn或RegularizerFn[]类型。

返回值：返回void。

示例 1：

Javascript

// Importing the tensorflow.js library
import * as tf from "@tensorflow/tfjs"
  
// Creating a model
const model = tf.sequential();
  
// Adding a layer
model.add(tf.layers.dense({units: 1, inputShape: [3]}));
  
// Defining input
const input = tf.tensor1d([1, 2, 3, 4]);
  
// Calling addLoss() method with its 
// parameter
const res = model.layers[0].addLoss([tf.abs(input)]);
  
// Printing output
console.log(JSON.stringify(input));
model.layers[0].getWeights()[0].print();

输出：

{"kept":false,"isDisposedInternal":false,"shape":[4],"dtype":"float32",
"size":4,"strides":[],"dataId":{"id":82},"id":124,"rankType":"1","scopeId":61}
Tensor
    [[0.143441  ],
     [-0.58002  ],
     [-0.5836995]]

示例 2：

Javascript

// Importing the tensorflow.js library
import * as tf from "@tensorflow/tfjs"
  
// Creating a model
const model = tf.sequential();
  
// Adding layers
model.add(tf.layers.dense({units: 1, inputShape: [3]}));
model.add(tf.layers.dense({units: 4}));
model.add(tf.layers.dense({units: 9, inputShape: [11]}));
  
// Defining inputs
const input1 = tf.tensor1d([0.5, 0.2, -33, null]);
const input2 = tf.tensor1d([0.33, 0.5, -1]);
const input3 = tf.tensor1d([1, 0.44]);
  
// Calling addLoss() method with its 
// parameter
const res1 = model.layers[0].addLoss([tf.cos(input1)]);
const res2 = model.layers[0].addLoss([tf.sin(input2)]);
const res3 = model.layers[0].addLoss([tf.tan(input3)]);
  
// Printing outputs
console.log(JSON.stringify(input1));
console.log(JSON.stringify(input2));
console.log(JSON.stringify(input3));
model.layers[0].getWeights()[0].print();

输出：

{"kept":false,"isDisposedInternal":false,"shape":[4],"dtype":"float32",
 "size":4,"strides":[],"dataId":{"id":169},"id":261,"rankType":"1","scopeId":112}
{"kept":false,"isDisposedInternal":false,"shape":[3],"dtype":"float32",
"size":3,"strides":[],"dataId":{"id":170},"id":262,"rankType":"1","scopeId":112}
{"kept":false,"isDisposedInternal":false,"shape":[2],"dtype":"float32",
"size":2,"strides":[],"dataId":{"id":171},"id":263,"rankType":"1","scopeId":112}
Tensor
    [[-0.0062826],
     [0.0883235 ],
     [-1.0633234]]

参考： https://js.tensorflow.org/api/latest/#tf.layers.Layer.addLoss