Open Advanced Search

Log in
Enterprise Plans
Free Trial

Log in
Free Trial
Enterprise Plans
Browse
Features
Pricing

DeepDyve

Advanced Search

show/hide query preview

With all of the words or DOIs:

DeepDyve's default query mode: search by keyword or DOI.

With the exact phrase:

Require these words, in this exact order. Do not surround your terms in double-quotes ("") in this field.

With at least one of the words:

Find any of these words, separated by spaces

Without the words:

Exclude each of these words, separated by spaces

With words in the article title:

Search for these terms only in the title of an article

Published between the years:

With the following authors:

Most effective as: LastName, First Name or Lastname, FN

With words in the journal name:

Search for articles published in journals where these words are in the journal name

Reset filters

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

DeepDyve requires Javascript to function. Please enable Javascript on your browser to continue.

Efficient Processing of Deep Neural Networks

Part of the Synthesis Lectures on Computer Architecture Book Series

DOI: 10.1007/978-3-031-01766-7

ISBN: 978-3-031-00638-8

Publisher: Springer International Publishing - Springer Journals

Jan 1, 2020

1 - 10 of 11 Chapters

Available

Introduction

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 3 –15

[Deep neural networks (DNNs) are currently the foundation for many modern artificial intelligence (AI) applications [5]. Since the breakthrough application of DNNs to speech recognition [6] and image recognition1 [7], the number of applications that use DNNs has exploded. These DNNs are employed...
Available

Overview of Deep Neural Networks

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 17 –39

[Deep Neural Networks (DNNs) come in a wide variety of shapes and sizes depending on the application.1 The popular shapes and sizes are also evolving rapidly to improve accuracy and efficiency. In all cases, the input to a DNN is a set of values representing the information to be analyzed by the...
Available

Key Metrics and Design Objectives

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 43 –58

[Over the past few years, there has been a significant amount of research on efficient processing of DNNs. Accordingly, it is important to discuss the key metrics that one should consider when comparing and evaluating the strengths and weaknesses of different designs and proposed techniques and...
Available

Kernel Computation

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 59 –72

[The fundamental computation of both CONV and FC layers described in Chapter 2 are multiply-and-accumulate (MAC) operations. Because there are negligible dependencies between these operations and the accumulations are commutative, there is considerable flexibility in the order in which MACs can...
Available

Designing DNN Accelerators

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 73 –118

[In Chapter 4, we discussed how DNN processing can undergo transforms to leverage optimized libraries or reduce the number of operations, specifically multiplications, in order to achieve higher performance (i.e., higher throughput and/or lower latency) on off-the-shelf general-purpose processors...
Available

Operation Mapping on Specialized Hardware

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 119 –143

[In Chapter 5, we discussed various key design considerations and techniques for the implementation of specialized DNN hardware. Also introduced was the notion of the mapping of the computation for a particular workload layer shape onto a specific DNN accelerator design, and the fact that the...
Available

Reducing Precision

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 147 –166

[As highlighted in the previous chapters, data movement dominates energy consumption and can affect the throughput for memory-bound systems. One way to address this issue to reduce the number of bits (bit width) required to represent the weights and activations of the DNN model. Using fewer bits...
Available

Exploiting Sparsity

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 167 –228

[A salient characteristic of the data used in DNN computations is that it is (or can be made to be) sparse. By saying that the data is sparse, we are referring to the fact that there are many repeated values in the data. Much of the time the repeated value is zero, which is what we will assume...
Available

Designing Efficient DNN Models

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 229 –252

[The previous two chapters discussed the use of DNN model and hardware co-design approaches, such as reducing precision (Chapter 7) and exploiting sparsity (Chapter 8), to reduce storage requirements, data movement, and the amount of MAC operations required for processing a DNN model. In this...
Available

Advanced Technologies

Sze, Vivienne; Chen, Yu-Hsin; Yang, Tien-Ju; Emer, Joel S.

Pages: 253 –280

[As highlighted throughout the previous chapters, data movement dominates energy consumption. The energy is consumed both in the access to the memory as well as the transfer of the data. The associated physical factors also limit the bandwidth available to deliver data between memory and compute,...

results per page

Browse All eBooks

Features

Search and discover articles on DeepDyve, PubMed, and Google Scholar
Read the full-text of open access and premium content
Organize articles with folders and bookmarks
Collaborate on and share articles and folders

Info

Pricing
Enterprise Plans
Browse Journals & Topics
About DeepDyve

Help

Help
Publishers
Contact Us

Sign up for your 14-Day Free Trial Now!

Read and print from thousands of top scholarly journals.

System error. Please try again!

Continue with Facebook

By signing up, you agree to DeepDyve’s Terms of Service and Privacy Policy.

Already have an account? Log in

Save Article to Bookmarks

Bookmark this article. You can see your Bookmarks on your DeepDyve Library.

To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.

Subscribe to Journal Email Alerts

To subscribe to email alerts, please log in first, or sign up for a DeepDyve account if you don’t already have one.

Follow a Journal

To get new article updates from a journal on your personalized homepage, please log in first, or sign up for a DeepDyve account if you don’t already have one.

Our policy towards the use of cookies

All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.

cookie policy