Generalization of an upper bound on the number of nodes needed to achieve linear separability

There is no fulltext present in this item.

Title:	Generalization of an upper bound on the number of nodes needed to achieve linear separability
Author(s):	Troost, M.; Seeliger, K. ; Gerven, M.A.J. van
Publication year:	2017
In:	Verheij, B.; Wiering, M. (ed.), BNAIC 2017: Preproceedings of the 29th Benelux Conference on Artificial Intelligence, pp. 238-252
Publisher:	Groningen : University of Groningen
ISBN:	9789403402994
Related links:	http://bnaic2017.ai.rug.nl/preproceedings.pdf
Annotation:	BNAIC 2017: The 29th Benelux Conference on Artificial Intelligence (Groningen, Netherlands, November 8-9, 2017)
Publication type:	Article in monograph or in proceedings
Please use this identifier to cite or link to this item : https://hdl.handle.net/2066/180887
Display more details
Editor(s):	Verheij, B.; Wiering, M.
Subject:	Cognitive artificial intelligence DI-BCB_DCC_Theme 4: Brain Networks and Neuronal Communication
Organization:	SW OZ DCC AI
Book title:	Verheij, B.; Wiering, M. (ed.), BNAIC 2017: Preproceedings of the 29th Benelux Conference on Artificial Intelligence
Page start:	p. 238
Page end:	p. 252
Abstract:	An important issue in neural network research is how to choose the number of nodes and layers such as to solve a classification problem. We provide new intuitions based on earlier results by [1] by deriving an upper bound on the number of nodes in networks with two hidden layers such that linear separability can be achieved. Concretely, we show that if the data can be described in terms of N finite sets and the used activation function f is non-constant, increasing and has a left asymptote, we can derive how many nodes are needed to linearly separate these sets. For the leaky rectified linear activation function, we prove separately that under some conditions on the slope, the same number of layers and nodes as for the aforementioned.
Languages used:	English (eng)