Getting Started With TFJS

Learn to make a deep learning model using Tensorflow Javascript. In this project, we'll build and train a deep learning model from scratch for the CIFAR-10 dataset. We'll also load a pre-trained model to make predictions on the test data.

You will learn to:

Make a deep learning model using TFJS

Train a model on CIFAR-10 dataset

Load a pre-trained model in TFJS

Use TensorBoard to visualize training

Skills

Machine Learning

Deep Learning

Deep Neural Networks

Prerequisites

Tensorflow

Javascript

Technologies

Node.js

TensorFlowJS

Project Description

Oftentimes, web developers need to use a machine learning model in a website. Instead of making a model in Python and connecting it with the front-end, we can make a model in Javascript and use it directly on the website. In this project, we’ll make a beginner-friendly deep learning model using TensorFlow Javascript.

In this project, we’ll use CIFAR-10 dataset. CIFAR-10 is a small dataset with 10 classes: airplane, automobile, bird, cat, deer, dog, frog, horse, ship, and truck. The classes are mutually exclusive. For example, the automobile class includes SUVs, sedans, and so on. On the other hand, truck class includes only large trucks.

After building and running the project, we’ll load a pre-trained model and use it to make predictions.

Project Tasks

Getting Started

Task 0: Instructions

Task 1: Loading Dataset

Preprocessing

Task 2: Normalization

Task 3: Reshape

Task 4: Preprocessing Features

Task 5: Implementing One-hot Encoding

Task 6: Preprocessing Labels

Building a Model

Task 7: Define the Model

Task 8: Compile Model

Task 9: Fit Model

Task 10: Start Training

Prediction

Task 11: Export Functions

Task 12: Load Trained Model

Task 13: Make Predictions

Congratulations!