Introduction to TensorFlow for Artificial Intelligence, Machine Learning, and Deep Learning

After taking the deep learning specialisation from deeplearning.ai last year I started the TensorFlow in Practice professional certificate now with this first introductory course.

Google AI advocate Laurence Moroney does an excellent job in smoothly introducing you into the world of AI with the market leading platform TensorFlow.

He starts with the "Hello world" version of TensorFlow, the most simple neural network, with one layer of one unit, to train a model on linear regression for data based on Y = 2X - 1, only 6 lines of code, including prediction.

Gradually more concepts are introduced and more complex models are build: computer vision on a 2 layered dense network, data normalization, activation functions, model evaluation, optimizers, loss functions, epochs and callbacks, adding convolutions and pooling, investigating the model summary, stochastic gradient descent, RMSProp, etc.

All comes with excellent examples in jupyter notebooks on the Google collaborator environment, no need to install anything on your computer. Grading is based on programming exercises (80%) and quizzes (20%).

This a far more easy introduction than the deep learning specialisation, with more focus on reading the data, building, training and evaluating the model, and experimenting with the hyperparameters and the network architecture. After finishing this course I felt competent to build my own deep learning networks.

I can recommend to jump into this course and this specialization after (or even without) a short introductory course in machine learning. No need to do the deep learning specialization first, you can do that afterwards if you want. There are link to some of Andrew Ng's course videos of deep learning in several of the modules for those who want to dive deeper or want to make their toes wet.