Modelli TensorFlow
TesorFlow.js
Una libreria JavaScript per
Formazione e distribuzione di
modelli di machine learning
nel browser
Modelli Tensorflow
Modelli e livelli sono elementi costitutivi importanti in Machine Learning .
Per diverse attività di Machine Learning è necessario combinare diversi tipi di livelli in un modello che può essere addestrato con i dati per prevedere i valori futuri.
TensorFlow.js supporta diversi tipi di modelli e diversi tipi di livelli.
Un modello TensorFlow è una rete neurale con uno o più livelli .
Un progetto Tensorflow
Un progetto Tensorflow ha questo flusso di lavoro tipico:
- Raccolta dati
- Creazione di un modello
- Aggiunta di livelli al modello
- Compilazione del modello
- Formazione del modello
- Utilizzando il modello
Esempio
Suppose you knew a function that defined a strait line:
Y = 1.2X + 5
Then you could calculate any y value with the JavaScript formula:
y = 1.2 * x + 5;
To demonstrate Tensorflow.js, we could train a Tensorflow.js model to predict Y values based on X inputs.
The TensorFlow model does not know the function.
// Create Training Data
const xs = tf.tensor([0, 1, 2, 3, 4]);
const ys = xs.mul(1.2).add(5);
// Define a Linear Regression Model
const model = tf.sequential();
model.add(tf.layers.dense({units:1, inputShape:[1]}));
// Specify Loss and Optimizer
model.compile({loss:'meanSquaredError', optimizer:'sgd'});
// Train the Model
model.fit(xs, ys, {epochs:500}).then(() => {myFunction()});
// Use the Model
function myFunction() {
const xArr = [];
const yArr = [];
for (let x = 0; x <= 10; x++) {
xArr.push(x);
let result = model.predict(tf.tensor([Number(x)]));
result.data().then(y => {
yArr.push(Number(y));
if (x == 10) {plot(xArr, yArr)};
});
}
}
The example is explained below:
Collecting Data
Create a tensor (xs) with 5 x values:
const xs = tf.tensor([0, 1, 2, 3, 4]);
Create a tensor (ys) with 5 correct y answers (multiply xs with 1.2 and add 5):
const ys = xs.mul(1.2).add(5);
Creating a Model
Create a sequential mode:.
const model = tf.sequential();
In a sequential model, the output from one layer is the input to the next layer.
Adding Layers
Add one dense layer to the model.
The layer is only one unit (tensor) and the shape is 1 (one dimentional):
model.add(tf.layers.dense({units:1, inputShape:[1]}));
in a dense the layer, every node is connected to every node in the preceding layer.
Compiling the Model
Compile the model using meanSquaredError as loss function and sgd (stochastic gradient descent) as optimizer function:
model.compile({loss:'meanSquaredError', optimizer:'sgd'});
Tensorflow Optimizers
- Adadelta -Implements the Adadelta algorithm.
- Adagrad - Implements the Adagrad algorithm.
- Adam - Implements the Adam algorithm.
- Adamax - Implements the Adamax algorithm.
- Ftrl - Implements the FTRL algorithm.
- Nadam - Implements the NAdam algorithm.
- Optimizer - Base class for Keras optimizers.
- RMSprop - Implements the RMSprop algorithm.
- SGD - Stochastic Gradient Descent Optimizer.
Training the Model
Train the model (using xs and ys) with 500 repeats (epochs):
model.fit(xs, ys, {epochs:500}).then(() => {myFunction()});
Using the Model
After the model is trained, you can use it for many different purposes.
This example predicts 10 y values, given 10 x values, and calls a function to plot the predictions in a graph:
function myFunction() {
const xArr = [];
const yArr = [];
for (let x = 0; x <= 10; x++) {
let result = model.predict(tf.tensor([Number(x)]));
result.data().then(y => {
xArr.push(x);
yArr.push(Number(y));
if (x == 10) {display(xArr, yArr)};
});
}
}
This example predicts 10 y values, given 10 x values, and calls a function to display the values:
function myFunction() {
const xArr = [];
const yArr = [];
for (let x = 0; x <= 10; x++) {
let result = model.predict(tf.tensor([Number(x)]));
result.data().then(y => {
xArr.push(x);
yArr.push(Number(y));
if (x == 10) {display(xArr, yArr)};
});
}
}