Sponsored by Amazon.
Understanding how to create a deep learning neural network is an essential component of any data scientist's knowledge base. This course covers some of the challenges that arise when training neural networks. It focuses on the problem of overfitting and its potential remedy: regularization. Learners should have a basic understanding of linear algebra and calculus.
- Discover what overfitting means and how to recognize it in deep learning models
- Understand how to sample your data to reduce the likelihood of overfitting
- Learn about regularization and its use as a remedy for overfitting
Laura Graesser is assisting in NVIDIA's autonomous driving project. Previously with The Boston Consulting Group, Laura is a graduate student at New York University, where she's working toward a master’s degree in computer science and machine learning. Laura's interests include neural networks and their application to computer vision problems, and the cross-fertilization between computer vision and natural language processing.
Table of Contents
- Introducing the Course 00:01:10
- What Is Overfitting? 00:03:43
- Regularization as a Remedy for Overfitting 00:03:31
- Reviewing Regularization Techniques—Early Stopping and Weight Regularization 00:08:42
- Reviewing Regularization Techniques—Dropout 00:07:18
- Improving the Way Neural Networks Learn 00:15:41
- Discussing Tips and Tricks for Training 00:07:11
- Title: Avoiding the Pitfalls of Deep Learning: Solving Model Overfitting with Regularization and Dropout
- Release date: November 2017
- Publisher(s): O'Reilly Media, Inc.
- ISBN: 9781491999622