What is TensorFlow? A Beginner's Guide to Know about it.

Reading Time:- 3 min 34 sec

Reading Time:- 3 min 34 sec

#tensorflow #what-is-tensorflow #machine-learning #deep-learning

Avatar

 BY TEJAS





TensorFlow is the end-to-end machine learning library for everyone. Anyone can use this openly. This is really the finest machine learning library in the world.

What is TensorFlow?

In today's world, we all know the importance of Deep Learning and Machine Learning. And for that, there is a famous library called TensorFlow. TensorFlow is the product of Google. Because it is the product of Google so of-course it is best. And other best part of Tensorflow is it is an open-source machine learning and deep learning library.


what is TensorFlow

What is TensorFlow



TensorFlow is developed by the Google Brain team for fast machine learning and deep learning.
TensorFlow is run on any devices and it has some wrappers in some languages like Python, Java, C++, Swift, and JavaScript. TensorFlow is also used in IOT platforms.


These are some companies that are using TensorFlow (According to Tensorflow Website).



ADVERTISEMENT




History of TensorFlow


TensorFlow's first stable version is released in 2017. As we have seen earlier TensorFlow is an open-source library. It is open-source under Apache open source license.



ADVERTISEMENT





How to run Tensorflow in Python and JavaScript?


For Python

To run TensorFlow in python we have to download TensorFlow's python module using pip.
For this use this command:

                   pip install tensorflow
                



For JavaScript


The best thing about TensorFlow is it run in javascript using the CDN link. This is quite interesting. To run TensorFlow in javascript we have to use the CDN link of TensorFlow.

                       
    <script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script>
                       
                   

You can install TensorFlow on different platforms please see TensorFlow's installation guide for how to install TensorFlow in different devices and frameworks?.



ADVERTISEMENT




How TensorFlow works?


So basically, TensorFlow works in three different processes.

  1. Processing the data
  2. Building the model
  3. Training and testing the model


Tensorflow takes the multi-dimensional arrays and process that arrays. Because it works on arrays and processes that array the name is TensorFlow. "Tensor" means arrays.
When we give some input into arrays it starts processing that array at one end and gives output on the second end.




ADVERTISEMENT





How to run Tensorflow in Python and JavaScript?


We are creating some simple programs of TensorFlow which also can be called the Hello-World program. The Programs we are covering in this article that are very basic. We see 2 programs one is for Python and the other is for javascript.
I used Google Colab for executing the example of Python.


For Python

First, we import TensorFlow using

                     
    import tensorflow as tf
                     
                 

Load the MNIST dataset in the program using

                     
    mnist = tf.keras.datasets.mnist
    (x_train, y_train), (x_test, y_test) = mnist.load_data()
                     
                 

Convert the sample data from integer to floating-point number by dividing it with 255.0

                     
    x_train, x_test = x_train / 255.0, x_test / 255.0
                     
                 

Build the model by stacking layers and choose the optimizer and loss function for training.

                     
    model = tf.keras.models.Sequential([
        tf.keras.layers.Flatten(input_shape=(28, 28)),
        tf.keras.layers.Dense(128, activation='relu'),
        tf.keras.layers.Dropout(0.2),
        tf.keras.layers.Dense(10)
    ])
                     
                 

for each example model returns a vector

                     
    predictions = model(x_train[:1]).numpy()
    predictions
                     
                 
Predictions

Predictions


tf.nn.softmax() function converts vector into probabilities.

                     
    tf.nn.softmax(predictions).numpy()
                     
                 
tf.nn.softmax()

tf.nn.softmax()


tf.keras.losses.SparseCategoricalCrossentropy() returns loss for each example.

                     
    loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)
                     
                 

model.compile() function compiles the model

                     
    model.compile(optimizer='adam',
                  loss=loss_fn,
                  metrics=['accuracy'])
                     
                 

model.fit() adjust the parameters to minimize the loss

                     
    model.fit(x_train, y_train, epochs=5)
                     
                 
model.fit()

model.fit()


model.evaluate() function checks the performance of the model

                     
    model.evaluate(x_test,  y_test, verbose=2)
                     
                 
model.evaluate()

model.evaluate()


Now time is to see the full program of TensorFlow in python is

                     
    # importing tensorflow
    import tensorflow as tf
    # load mnist dataset
    mnist = tf.keras.datasets.mnist
    (x_train, y_train), (x_test, y_test) = mnist.load_data()
    # convert integers into float
    x_train, x_test = x_train / 255.0, x_test / 255.0
    # building the model
    model = tf.keras.models.Sequential([
        tf.keras.layers.Flatten(input_shape=(28, 28)),
        tf.keras.layers.Dense(128, activation='relu'),
        tf.keras.layers.Dropout(0.2),
        tf.keras.layers.Dense(10)
    ])
    # return a vector
    predictions = model(x_train[:1]).numpy()
    predictions
    # converting into probabilities
    tf.nn.softmax(predictions).numpy()
    # returns loss 
    loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)
    # compiling the model
    model.compile(optimizer='adam',
              loss=loss_fn,
              metrics=['accuracy'])
    # fitting model
    model.fit(x_train, y_train, epochs=5)
    # check models performance
    model.evaluate(x_test,  y_test, verbose=2)
                     
                 

In this model we get a loss is nearly about ~7% and accuracy is about ~97%

loss and accuracy of model


ADVERTISEMENT




For JavaScript


For JavaScript, there are some different functions in TensorFlow. The output of this program is shown in the console of the browser and also in the <H1> tag.

First, we are adding the CDN link of TensorFlow in the <head> tag. After the addition of the link now we have created an array and inserted fifteen random values into the array and declared it into a values variable.
Then we have created a second array in which there are two values in that array [5, 3] and declared it into a shape variable.

And after that, we create a data variable and in that variable, we pass the tf.tensor(values, shapes) function. and then in last, we use console.log() for printing the output into the console.
And data.toString() function converts data into a string and displays the string values in the console.
And by using document.getElementById("demo").innerHTML = data.toString(); we show output in <H1>> tag (It is just a basic HTML and JavaScript).


This is console output of program.

Console Output

Console Output


And this is output of program by using <H1> tag.

Page Output

Page Output


So this is full HTML and JavaScript code for TensorFlow in JavaScript example.

                    
    <!DOCTYPE html>
    <head>
        <script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script>
    </head>
    <body>
        <h1 id="demo"></h1>
        <script>
            const values = [10.5, 6.5, 55.5, 96.96, 36.25, 12.25, 87.00, 35.65, 33.96, 25.14, 8.00, 9.66, 22.12, 66.20, 45.33];
            const shape = [5, 3];
            const data = tf.tensor(values, shape);
            console.log(data.toString());
            document.getElementById("demo").innerHTML = data.toString();
        </script>
    </body>
    </html>
                    
                

The following videos will inspire you to learn artificial intelligence. These videos are on TensorFlow's official YouTube channel.

Videos by TensorFlow


ADVERTISEMENT




And we're done here


Conclusion:-

We have covered in this article "What is TensorFlow?", "History of TensorFlow", "How to run Tensorflow in Python and Javascript?", "How TensorFlow works?", "A simple programs in TensorFlow (For Python and JavaScript)".
If You like this article, please comment and share. also, you have any doubts about TensorFlow then please comment below.

Thank you.


ADVERTISEMENT







Search
Advertisement
Advertisement
Share With
Follow us on
Advertisement
Advertisement