Analog versus discrete neural networks

Authors:
Bhaskar DasGupta;Georg Schnitger
Affiliations:
Department of Computer Science, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada;Fachbereich 20, Informatik, Universitat Frankfurt, 60054 Frankfurt, Germany
Venue:
Neural Computation
Year:
1996

Citing 8
Cited 5

The complexity of Boolean functions

The complexity of Boolean functions
The complexity of finite functions

Handbook of theoretical computer science (vol. A)
On the computational power of sigmoid versus boolean threshold circuits (extended abstract)

SFCS '91 Proceedings of the 32nd annual symposium on Foundations of computer science
The complexity of the parity function in unbounded fan-in, unbounded depth circuits

Theoretical Computer Science
Computational limitations on training sigmoid neural networks

Information Processing Letters
Finiteness results for sigmoidal “neural” networks

STOC '93 Proceedings of the twenty-fifth annual ACM symposium on Theory of computing
Bounds for the computational power and learning complexity of analog neural nets

STOC '93 Proceedings of the twenty-fifth annual ACM symposium on Theory of computing
The Power of Approximation: A Comparison of Activation Functions

Advances in Neural Information Processing Systems 5, [NIPS Conference]

The Connectionist Inductive Learning and Logic Programming System

Applied Intelligence
Neural circuits for pattern recognition with small total wire length

Theoretical Computer Science - Natural computing
General-purpose computation with neural networks: a survey of complexity theoretic results

Neural Computation
Spiking neural controllers for pushing objects around

SAB'06 Proceedings of the 9th international conference on From Animals to Animats: simulation of Adaptive Behavior
Assessing the effectiveness of artificial neural networks on problems related to elliptic curve cryptography

Mathematical and Computer Modelling: An International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

We show that neural networks with three-times continuously differentiable activation functions are capable of computing a certain family of n-bit Boolean functions with two gates, whereas networks composed of binary threshold functions require at least â聞娄(log n) gates. Thus, for a large class of activation functions, analog neural networks can be more powerful than discrete neural networks, even when computing Boolean functions.