Publication Details

TypeCNN: CNN Development Framework With Flexible Data Types

REK Petr and SEKANINA Lukáš. TypeCNN: CNN Development Framework With Flexible Data Types. In: Design, Automation and Test in Europe Conference. Florence: European Design and Automation Association, 2019, pp. 292-295. ISBN 978-3-9819263-2-3.
Czech title
TypeCNN: Vývojové prostředí pro CNN s flexibilními datovými typy
Type
conference paper
Language
english
Authors
Rek Petr, Ing. (FIT BUT)
Sekanina Lukáš, prof. Ing., Ph.D. (DCSY FIT BUT)
URL
Keywords
convolutional neural network, software library, data type, deep learning
Abstract
The rapid progress in artificial intelligence technologies based on deep and convolutional neural networks (CNN) has led to an enormous interest in efficient implementations of neural networks in embedded devices and hardware. We present a new software framework for the development of (approximate) convolutional neural networks in which the user can define and use various data types for forward (inference) procedure, backward (training) procedure and weights. Moreover, non-standard arithmetic operations such as approximate multipliers can easily be integrated into the CNN under design. This flexibility enables to analyze the impact of chosen data types and non-standard arithmetic operations on CNN training and inference efficiency. The framework was implemented in C++ and evaluated using several case studies. 
Published
2019
Pages
292-295
Proceedings
Design, Automation and Test in Europe Conference
Conference
Design, Automation and Test in Europe Conference, Florencie, IT
ISBN
978-3-9819263-2-3
Publisher
European Design and Automation Association
Place
Florence, IT
BibTeX
@INPROCEEDINGS{FITPUB11854,
   author = "Petr Rek and Luk\'{a}\v{s} Sekanina",
   title = "TypeCNN: CNN Development Framework With Flexible Data Types",
   pages = "292--295",
   booktitle = "Design, Automation and Test in Europe Conference",
   year = 2019,
   location = "Florence, IT",
   publisher = "European Design and Automation Association",
   ISBN = "978-3-9819263-2-3",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11854"
}
Files
Back to top