Online Course – Certified Professional Specialization in Machine Learning from Imperial College London

Getting started with TensorFlow 2

• Course 1 • 26 hours • 4.9 (567 ratings)

Course Details

What you’ll learn

Welcome to the “Getting Started with TensorFlow 2” course! In this course, you will learn the entire process for developing deep learning models with TensorFlow, from characterization, training, evaluation, and prediction with models via the serial API, model validation, incorporating regularization, implementing callbacks, and saving and loading models.

Apply the concepts you learn immediately with hands-on coding exercises, guided by a certified teaching assistant. Plus, there’s a series of automatically assessed programming assignments to strengthen your skills.

At the end of the course, you will bring together the concepts in a final project, where you will develop an image classifier model from scratch.

TensorFlow is an open source machine learning library and one of the most widely used frameworks for deep learning. The release of TensorFlow 2 marks a significant shift in product development, with a primary focus on ease of use for all users, from beginners to advanced. This course is designed for both complete beginners and those with experience with TensorFlow 1.x.

The knowledge required to succeed in the course is:

• Proficiency in the Python programming language (the course uses Python 3)
• Knowledge of general machine learning concepts (such as overfitting and underfitting, supervised learning tasks, validation, regularization, and model selection)
• Training in the field of deep learning, including typical model architectures (MLP, convolutional neural networks), activation functions, output layers, and optimization.

To customize models with TensorFlow 2

• Course 2 • 27 hours • 4.8 (188 ratings)

Course Details

What you’ll learn

Welcome to the “Customizing Your Models with TensorFlow 2” course! In this course, you will deepen your knowledge and skills in TensorFlow to develop custom deep learning models and techniques for any application. You will use low-level TensorFlow APIs to develop complex model architectures, custom layers, and flexible data. You will also expand your knowledge of TensorFlow APIs to include series models.

Apply concepts immediately with hands-on exercises, guided by a certified teaching assistant. Plus, there’s a series of automatically assessed programming assignments to strengthen your skills.

At the end of the course, you will group the concepts into a final project, in which you will develop a custom neural translation model from scratch.

TensorFlow is an open source machine learning library and one of the most widely used frameworks in deep learning. The release of TensorFlow 2 marks a significant shift in product development, with a primary focus on ease of use for all users, from beginners to advanced.

This course follows directly on from the previous course “Getting Started with TensorFlow 2”. The additional knowledge required to succeed is:

• Proficiency in the Python programming language (the course uses Python 3)
• General knowledge of machine learning concepts (such as overfitting and underfitting, supervised learning tasks, validation, regularization, and model selection)
• Training in the field of deep learning, including typical model architectures (MLP, CNN, RNN, ResNet) and concepts such as transfer learning, data augmentation, and saliency walk.

Probability-driven information with TensorFlow 2

• Course 3 • 52 hours • 4.7 (101 ratings)

Course Details

What you’ll learn

Welcome to the “Probabilistic Information with TensorFlow” course! This course builds on the basic TensorFlow concepts and skills learned in the first two courses, and focuses on a probabilistic approach to deep learning. This is a very important area that aims to quantify the noise and uncertainty often found in real-world data. This is a cardinal aspect when using deep learning models in areas such as autonomous vehicles or medical diagnostics; it is important for the model to know what it does not know.

You will learn how to develop probabilistic models with TensorFlow, using the TensorFlow Probability Library, which is designed to make it easier to integrate probabilistic models with deep learning. Thus, this course can also be considered an introduction to the TensorFlow Probability Library.

You will learn how probabilistic densities can be represented and incorporated into deep learning models in TensorFlow, including Bayesian neural networks, normal streams, and variational autoencoders. You will learn to develop models for quantifying uncertainty, as well as generative models that can generate new examples similar to those in the data, such as images of celebrity faces.

Apply concepts through hands-on exercises, guided by a certified teaching assistant. In addition, there is a series of automatically assessed programming assignments to strengthen your skills.

At the end of the course, you will group the concepts into a final project, in which you will develop a variable autoencoder algorithm to create a generative model of a set of synthetic images that you will need to create yourself.

This course follows the two previous courses in the specialization, “Getting Started with TensorFlow 2” and “Tuning Your Models with TensorFlow 2.” Additional knowledge required to succeed is:

• A solid foundation in probability and statistics
• Good knowledge of standard probability distributions, probability density functions, and concepts such as maximum likelihood estimates, the change of variables formula for random variables, and the lower bound of evidence (ELBO) used in variational inference.