Learning Deep CNN Structures

Konferenz: CNNA 2016 - 15th International Workshop on Cellular Nanoscale Networks and their Applications
23.08.2016 - 25.08.2016 in Dresden, Deutschland

Tagungsband: CNNA 2016

Seiten: 2Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Autoren:
Mueller, Jens; Walz, Simon; Tetzlaff, Ronald (Institute of Principles of Electrical and Electronic Engineering, Faculty of Electrical and Computer Engineering, Technische Universität Dresden, 01062 Dresden, Germany)

Inhalt:
Multi-layer CNN with a large number of layers (socalled deep CNN structures) are considered in this paper. A backpropagation algorithm is used in order to find the network parameters for a sequence of operations. The proposed learning method is able to train and optimise CNN programs that are suitable for the implementation on state-of-the-art digital designs.