Loved by learners at thousands of companies
Discover Deep Learning ApplicationsDeep learning is the machine learning technique behind the most exciting capabilities in robotics, natural language processing, image recognition, and artificial intelligence. In this 4-hour course, you’ll gain hands-on practical knowledge of how to apply your Python skills to deep learning with the Keras 2.0 library.
Explore Keras Models with a Library ContributorTaught by ex-Google data scientist and Keras contributor, Dan Becker, this deep learning course explores neural network models and how you can generate predictions with them. The first chapters will grow your understanding of both forward and backward propagation and how they work in practice.
Keras library is a Python library that can help you develop and review deep learning models. Like many Python libraries, it's free, open-source and very user friendly. You’ll start by creating a Keras model and will learn how to compile, fit, and classify it before making predictions. Once you’ve completed this course, you’ll have all the tools you need to build deep neural networks and start experimenting with wider and deeper networks over time.
Delve Further into Deep LearningThis course is part of several machine learning and deep learning tracks, offering you clear pathways to build your skills and experience in this area once you’ve completed the introductory course, whether you want to complete a personal project or move towards a career as a Machine Learning Scientist.
Basics of deep learning and neural networksFree
In this chapter, you'll become familiar with the fundamental concepts and terminology used in deep learning, and understand why deep learning techniques are so powerful today. You'll build simple neural networks and generate predictions with them.Introduction to deep learning50 xpComparing neural network models to classical regression models50 xpForward propagation50 xpCoding the forward propagation algorithm100 xpActivation functions50 xpThe Rectified Linear Activation Function100 xpApplying the network to many observations/rows of data100 xpDeeper networks50 xpForward propagation in a deeper network50 xpMulti-layer neural networks100 xpRepresentations are learned50 xpLevels of representation50 xp
Optimizing a neural network with backward propagation
Learn how to optimize the predictions generated by your neural networks. You'll use a method called backward propagation, which is one of the most important techniques in deep learning. Understanding how it works will give you a strong foundation to build on in the second half of the course.The need for optimization50 xpCalculating model errors50 xpUnderstanding how weights change model accuracy50 xpCoding how weight changes affect accuracy100 xpScaling up to multiple data points100 xpGradient descent50 xpCalculating slopes100 xpImproving model weights100 xpMaking multiple updates to weights100 xpBackpropagation50 xpThe relationship between forward and backward propagation50 xpThinking about backward propagation50 xpBackpropagation in practice50 xpA round of backpropagation50 xp
Building deep learning models with keras
In this chapter, you'll use the Keras library to build deep learning models for both regression and classification. You'll learn about the Specify-Compile-Fit workflow that you can use to make predictions, and by the end of the chapter, you'll have all the tools necessary to build deep neural networks.Creating a Keras model50 xpUnderstanding your data50 xpSpecifying a model100 xpCompiling and fitting a model50 xpCompiling the model100 xpFitting the model100 xpClassification models50 xpUnderstanding your classification data50 xpLast steps in classification models100 xpUsing models50 xpMaking predictions100 xp
Fine-tuning keras models
Learn how to optimize your deep learning models in Keras. Start by learning how to validate your models, then understand the concept of model capacity, and finally, experiment with wider and deeper networks.Understanding model optimization50 xpDiagnosing optimization problems50 xpChanging optimization parameters100 xpModel validation50 xpEvaluating model accuracy on validation dataset100 xpEarly stopping: Optimizing the optimization100 xpExperimenting with wider networks100 xpAdding layers to a network100 xpThinking about model capacity50 xpExperimenting with model structures50 xpStepping up to images50 xpBuilding your own digit recognition model100 xpFinal thoughts50 xp
PrerequisitesSupervised Learning with scikit-learn
Data Scientist and contributor to Keras and TensorFlow libraries
Dan Becker is a data scientist with years of deep learning experience. He has contributed to the Keras and TensorFlow libraries, finishing 2nd (out of 1353 teams) in the $3million Heritage Health Prize competition, and supervised consulting projects for 6 companies in the Fortunate 100. He previously worked as a data scientist at Google. Now he is the CEO and co-founder of Decision.ai, which helps companies apply their machine learning models to make better real-world decisions.