Efficient Deep Learning (First Early Release)

Make deep learning models more efficient by using fewer resources to deliver better quality. With this book, you'll examine algorithms and techniques used by researchers and engineers at Google, Facebook, and other AI labs to train and deploy models on devices ranging from server-side machines to tiny microcontrollers.

Gaurav Menghani, staff software engineer at Google Research, examines the fundamentals and practical techniques that can help you optimize your model training and deployment workflows. Your models will perform as well as or better than before using a fraction of resources.

If you're an engineer, research scientist, or student interested in training and deploying models to production, you'll dive deep into techniques and infrastructure before getting your hands dirty with practical projects. This report helps you:

• Examine the state of deep learning, its applications, and its rapid growth
• Optimize existing models for efficient training and deployment
• Learn the latest tools and techniques to achieve model optimization
• Apply compression and optimization techniques to new models
• Explore how the state-of-the-art in-model efficiency applies to your problem
• Understand trade-offs between model footprint and quality and determine what works best in your case