Vishal News Web

Simple Neural Network with TensorFlow.js

async function trainModel() { // Create a sequential model const model = tf.sequential(); // Add a dense layer with 20 input units and ReLU activation model.add(tf.layers.dense({ inputShape: [20], units: 10, activation: 'relu' })); // Add another dense layer with 3 units and softmax activation model.add(tf.layers.dense({ units: 3, activation: 'softmax' })); // Compile the model with a categorical crossentropy loss function and SGD optimizer model.compile({ optimizer: tf.train.sgd(0.1), loss: 'categoricalCrossentropy' }); // Generate sample data: 3 input samples with 20 features each const xs = tf.tensor2d([ Array.from({ length: 20 }, () => Math.random()), Array.from({ length: 20 }, () => Math.random()), Array.from({ length: 20 }, () => Math.random()) ]); // Generate sample target data: one-hot encoded labels const ys = tf.tensor2d([[1, 0, 0], [0, 1, 0], [0, 0, 1]]); // Train the model await model.fit(xs, ys, { epochs: 50, callbacks: { onEpochEnd: (epoch, logs) => { console.log(`Epoch ${epoch + 1}: loss = ${logs.loss}`); } } }); // Make predictions const prediction = model.predict(xs); prediction.print(); // Display predictions const outputDiv = document.getElementById('output'); const predictionsArray = await prediction.array(); predictionsArray.forEach((pred, index) => { const predDiv = document.createElement('div'); predDiv.classList.add('prediction', index % 2 === 0 ? 'red' : 'blue'); predDiv.textContent = `Sample ${index + 1} Prediction: ${pred}`; outputDiv.appendChild(predDiv); }); } window.onload = trainModel;