📜  Tensorflow.js tf.train.momentum()函数

📅  最后修改于: 2022-05-13 01:56:49.694000             🧑  作者: Mango

Tensorflow.js tf.train.momentum()函数

Tensorflow.js 是谷歌开发的一个开源库,用于在浏览器或节点环境中运行机器学习模型和深度学习神经网络。

tf.train.momemtum()函数用于创建使用动量梯度下降算法的 tf.MomentumOptimizer。

句法:

tf.train.momentum(learningRate, momentum, useNesterov)

参数:

  • learningRate(数字):它指定将被动量梯度下降算法使用的学习率。
  • 动量(数字):它指定动量梯度下降算法将使用的动量。
  • useNesterov (boolean):指定是否使用nesterov 动量。它是一个可选参数。

返回值:它返回一个 tf.MomentumOptimizer

示例1:通过学习系数 a 和 b,使用动量优化器拟合函数f=(a*x+b)。在这个例子中,我们将使用nesterov 动量。所以 useNestrov 将是真的。

Javascript
// Importing tensorflow
import * as tf from "@tensorflow/tfjs"
  
const xs = tf.tensor1d([0, 1, 2]);
const ys = tf.tensor1d([1.1, 5.9, 16.8]);
  
const a = tf.scalar(Math.random()).variable();
const b = tf.scalar(Math.random()).variable();
  
const f = x => a.mul(x).add(b);
const loss = (pred, label) => pred.sub(label).square().mean();
  
const learningRate = 0.01;
const momentum = 10;
const useNestrov = true;
const optimizer = tf.train.momentum(learningRate, momentum, useNestrov);
  
// Train the model.
for (let i = 0; i < 10; i++) {
   optimizer.minimize(() => loss(f(xs), ys));
}
  
// Make predictions.
console.log(
     `a: ${a.dataSync()}, b: ${b.dataSync()}}`);
const preds = f(xs).dataSync();
preds.forEach((pred, i) => {
   console.log(`x: ${i}, pred: ${pred}`);
});


Javascript
// Importing tensorflow
import * as tf from "@tensorflow/tfjs"
  
const xs = tf.tensor1d([0, 1, 2, 3]);
const ys = tf.tensor1d([1.1, 5.9, 16.8, 33.9]);
  
const a = tf.scalar(Math.random()).variable();
const b = tf.scalar(Math.random()).variable();
const c = tf.scalar(Math.random()).variable();
  
const f = x => a.mul(x.square()).add(b.mul(x)).add(c);
const loss = (pred, label) => pred.sub(label).square().mean();
  
const learningRate = 0.01;
const momentum = 10;
const useNestrov = false;
const optimizer = tf.train.momentum(learningRate, momentum, useNestrov);
  
// Train the model.
for (let i = 0; i < 10; i++) {
   optimizer.minimize(() => loss(f(xs), ys));
}
  
// Make predictions.
console.log(
     `a: ${a.dataSync()}, b: ${b.dataSync()}, c: ${c.dataSync()}`);
const preds = f(xs).dataSync();
preds.forEach((pred, i) => {
   console.log(`x: ${i}, pred: ${pred}`);
});


输出:

a: 1982014720, b:1076448384
x: 0, pred: 1076448384
x: 1, pred: 3058463232
x: 2, pred: 5040477696

示例2:通过学习系数 a 和 b,使用动量优化器拟合二次方程。在这个例子中,我们不会使用nesterov 动量。所以 useNestrov 将是错误的。

Javascript

// Importing tensorflow
import * as tf from "@tensorflow/tfjs"
  
const xs = tf.tensor1d([0, 1, 2, 3]);
const ys = tf.tensor1d([1.1, 5.9, 16.8, 33.9]);
  
const a = tf.scalar(Math.random()).variable();
const b = tf.scalar(Math.random()).variable();
const c = tf.scalar(Math.random()).variable();
  
const f = x => a.mul(x.square()).add(b.mul(x)).add(c);
const loss = (pred, label) => pred.sub(label).square().mean();
  
const learningRate = 0.01;
const momentum = 10;
const useNestrov = false;
const optimizer = tf.train.momentum(learningRate, momentum, useNestrov);
  
// Train the model.
for (let i = 0; i < 10; i++) {
   optimizer.minimize(() => loss(f(xs), ys));
}
  
// Make predictions.
console.log(
     `a: ${a.dataSync()}, b: ${b.dataSync()}, c: ${c.dataSync()}`);
const preds = f(xs).dataSync();
preds.forEach((pred, i) => {
   console.log(`x: ${i}, pred: ${pred}`);
});

输出:

a: 892235776, b: 331963616, c: 134188384
x:0, pred: 134188384
x:1, pred: 1358387840
x:2, pred: 4367058944
x:3, pred: 9160201216

参考: https://js.tensorflow.org/api/1.0.0/#train.momentum