Search icon CANCEL
Subscription
0
Cart icon
Your Cart (0 item)
Close icon
You have no products in your basket yet
Arrow left icon
Explore Products
Best Sellers
New Releases
Books
Videos
Audiobooks
Learning Hub
Free Learning
Arrow right icon
Arrow up icon
GO TO TOP
Hands-On Meta Learning with Python

You're reading from   Hands-On Meta Learning with Python Meta learning using one-shot learning, MAML, Reptile, and Meta-SGD with TensorFlow

Arrow left icon
Product type Paperback
Published in Dec 2018
Publisher Packt
ISBN-13 9781789534207
Length 226 pages
Edition 1st Edition
Languages
Arrow right icon
Author (1):
Arrow left icon
Sudharsan Ravichandiran Sudharsan Ravichandiran
Author Profile Icon Sudharsan Ravichandiran
Sudharsan Ravichandiran
Arrow right icon
View More author details
Toc

Table of Contents (12) Chapters Close

Preface 1. Introduction to Meta Learning 2. Face and Audio Recognition Using Siamese Networks FREE CHAPTER 3. Prototypical Networks and Their Variants 4. Relation and Matching Networks Using TensorFlow 5. Memory-Augmented Neural Networks 6. MAML and Its Variants 7. Meta-SGD and Reptile 8. Gradient Agreement as an Optimization Objective 9. Recent Advancements and Next Steps 10. Assessments 11. Other Books You May Enjoy

Memory-augmented neural networks (MANN)

Now we will see an interesting variant of NTM, called MANN. It is extensively used for one-shot learning tasks. MANN is designed to make NTM perform better at one-shot learning tasks. We know that NTM can either use content-based addressing or location-based addressing. But in MANN, we use only content-based addressing.

MANN uses a new addressing scheme called least recently used access. As the name suggests, it writes to the least recently used memory location. Wait. What? We just learned that MANN is not location-based, so why are we writing to the least recently used location? This is because the least recently used memory location is determined by the read operation and the read operation is performed by content-based addressing. So, we basically perform content-based addressing for reading and write to the location that was least recently...

lock icon The rest of the chapter is locked
Register for a free Packt account to unlock a world of extra content!
A free Packt account unlocks extra newsletters, articles, discounted offers, and much more. Start advancing your knowledge today.
Unlock this book and the full library FREE for 7 days
Get unlimited access to 7000+ expert-authored eBooks and videos courses covering every tech area you can think of
Renews at $19.99/month. Cancel anytime
Banner background image