Tensorflow.js is an open-source library developed by Google for running machine learning models and deep learning neural networks in the browser or node environment. The tf.initializers.Initializer() class is used to extend serialization.Serializable class. It is the base class of Initializer.
This tf.initializers.Initializer class contains fifteen inbuilt functions which are illustrated below:
- tf.initializers.Initializer class .constant() function
- tf.initializers.Initializer class .glorotNormal() function
- tf.initializers.Initializer class .glorotUniform() function
- tf.initializers.Initializer class .heNormal() function
- tf.initializers.Initializer class .heUniform() function
- tf.initializers.Initializer class .identity() function
- tf.initializers.Initializer class .leCunNormal() function
- tf.initializers.Initializer class .leCunUniform() function
- tf.initializers.Initializer class .ones() function
- tf.initializers.Initializer class .orthogonal() function
- tf.initializers.Initializer class .randomNormal() function
- tf.initializers.Initializer class .randomUniform() function
- tf.initializers.Initializer class .truncatedNormal() function
- tf.initializers.Initializer class .varianceScaling() function
- tf.initializers.Initializer class .zeros() function
1. tf.initializers.Initializer class .constant() function: It is used to generate the values initialized to some constant.
Example:
Javascript
const tf = require("@tensorflow/tfjs")
var initializer = tf.initializers.constant({ value: 7, })
console.log(initializer);
|
Output:
Constant { value: 7 }
2. tf.initializers.Initializer class .glorotNormal() function: It extract samples from a truncated normal distribution which is been centered at 0 with stddev = sqrt(2 / (fan_in + fan_out)). Note, that the fan_in is the number of inputs in the tensor weight and the fan_out is the number of outputs in the tensor weight.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
console.log(tf.initializers.glorotNormal(9));
console.log('\nIndividual values:\n');
console.log(tf.initializers.glorotNormal(9).scale);
console.log(tf.initializers.glorotNormal(9).mode);
console.log(tf.initializers.glorotNormal(9).distribution);
|
Output:
{
"scale": 1,
"mode": "fanAvg",
"distribution": "normal"
}
Individual values:
1
fanAvg
normal
3. tf.initializers.Initializer class .glorotUniform() function: It is used to extract samples from a uniform distribution within [-limit, limit] where limit is sqrt(6 / (fan_in + fan_out)) where fan_in is the number of input units in the weight tensor and fan out is the number of output units in the weight tensor.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
const geek = tf.initializers.glorotUniform(7)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.scale);
console.log(geek.mode);
console.log(geek.distribution);
|
Output:
{
"scale": 1,
"mode": "fanAvg",
"distribution": "uniform"
}
Individual values:
1
fanAvg
uniform
4. tf.initializers.Initializer class .heNormal() function: It is used to draw samples from a truncated normal distribution centered on zero with stddev = sqrt(2 / fanIn) within [-limit, limit] where, limit is sqrt(6 / fan_in). Note, that the fanIn is the number of inputs in the tensor weight.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
const geek = tf.initializers.heNormal(7)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.scale);
console.log(geek.mode);
console.log(geek.distribution);
|
Output:
{
"scale": 2,
"mode": "fanIn",
"distribution": "normal"
}
Individual values:
2
fanIn
normal
5. tf.initializers.Initializer class .heUniform() function: It draws samples from a uniform distribution within [-cap, cap] where, cap is sqrt(6 / fan_in). Note, that the fanIn is the number of inputs in the tensor weight.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
const geek = tf.initializers.heUniform(7)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.scale);
console.log(geek.mode);
console.log(geek.distribution);
|
Output:
{
"scale": 2,
"mode": "fanIn",
"distribution": "uniform"
}
Individual values:
2
fanIn
uniform
6. tf.initializers.Initializer class .identity() function: It is used to return a new tensor object with an identity matrix. Its only used for 2D matrices.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
const value=tf.initializers.identity(1.0)
console.log(value)
|
Output:
{
"gain": 1
}
7. tf.initializers.Initializer class .leCunNormal() function: It is used to extract samples from a truncated normal distribution which is centered at zero with stddev = sqrt(1 / fanIn). Note, that fanIn is the number of inputs in the tensor weight.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
const geek = tf.initializers.leCunNormal(3)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.scale);
console.log(geek.mode);
console.log(geek.distribution);
|
Output:
{
"scale": 1,
"mode": "fanIn",
"distribution": "normal"
}
Individual values:
1
fanIn
normal
8. tf.initializers.Initializer class .leCunUniform() function: It takes samples from a uniform distribution in the interval [-cap, cap] with cap = sqrt(3 / fanIn). Note, that fanIn is the number of inputs in the tensor weight.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
console.log(tf.initializers.leCunUniform(4));
console.log("\nIndividual Values\n");
console.log(tf.initializers.leCunUniform(4).scale);
console.log(tf.initializers.leCunUniform(4).mode);
console.log(tf.initializers.leCunUniform(4).distribution);
|
Output:
{
"scale": 1,
"mode": "fanIn",
"distribution": "uniform"
}
Individual Values
1
fanIn
uniform
9. tf.initializers.Initializer class .ones() function: It is used to create a tensor with all elements set to 1, or it initializes tensor with value 1.
Example:
Javascript
const tf=require("@tensorflow/tfjs")
var GFG=tf.ones([3, 4]);
GFG.print()
|
Output:
Tensor
[[1, 1, 1, 1],
[1, 1, 1, 1],
[1, 1, 1, 1]]
10. tf.initializers.Initializer class .orthogonal() function: It produces a random orthogonal matrix.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
let geek = tf.initializers.orthogonal(2)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.DEFAULT_GAIN);
console.log(geek.gain);
|
Output:
{
"DEFAULT_GAIN": 1,
"gain": 1
}
Individual values:
1
1
11. tf.initializers.Initializer class .randomNormal() function: It is used to produce random values that are initialized to a normal distribution.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
let geek = tf.initializers.randomNormal(3)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.DEFAULT_MEAN);
console.log(geek.DEFAULT_STDDEV);
console.log(geek.mean);
console.log(geek.stddev);
|
Output:
{
"DEFAULT_MEAN": 0,
"DEFAULT_STDDEV": 0.05,
"mean": 0,
"stddev": 0.05
}
Individual values:
0
0.05
0
0.05
12. tf.initializers.Initializer class .randomUniform() function: It is used to generate random values that are initialized to a uniform distribution. The values are distributed uniformly between the configured min-value and max-value.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
let geek = tf.initializers.randomUniform(5)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.DEFAULT_MINVAL);
console.log(geek.DEFAULT_MAXVAL);
console.log(geek.minval);
console.log(geek.maxval);
|
Output:
{
"DEFAULT_MINVAL": -0.05,
"DEFAULT_MAXVAL": 0.05,
"minval": -0.05,
"maxval": 0.05
}
Individual values:
-0.05
0.05
-0.05
0.05
13. tf.initializers.Initializer class .truncatedNormal(): It function produces random values initialized to a truncated normal distribution.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
let geek = tf.initializers.truncatedNormal(13)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.DEFAULT_MEAN);
console.log(geek.DEFAULT_STDDEV);
console.log(geek.mean);
console.log(geek.stddev);
|
Output:
{
"DEFAULT_MEAN": 0,
"DEFAULT_STDDEV": 0.05,
"mean": 0,
"stddev": 0.05
}
Individual values:
0
0.05
0
0.05
14. tf.initializers.Initializer class .varianceScaling() function: It is capable of adjusting its scale to the shape of weights. Using the value of distribution = NORMAL, samples are drawn from a truncated normal distribution that has center at 0, with stddev = sqrt(scale / n).
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
let geek = tf.initializers.varianceScaling(33)
console.log(geek);
console.log('\nIndividual values:\n');
console.log(geek.scale);
console.log(geek.mode);
console.log(geek.distribution);
|
Output:
{
"scale": 1,
"mode": "fanIn",
"distribution": "normal"
}
Individual values:
1
fanIn
normal
15. tf.initializers.Initializer class .zeros() function: It is an initializer that is used to produce tensors that are initialized to zero.
Example:
Javascript
import * as tf from "@tensorflow/tfjs"
const initializer = tf.initializers.zeros();
console.log(JSON.stringify(+initializer));
|
Output:
null
Reference: https://js.tensorflow.org/api/latest/#class:initializers.Initializer
Like Article
Suggest improvement
Share your thoughts in the comments
Please Login to comment...