Abstract
The use of prior knowledge to train neural networks for better performance has attracted increased attention. Initial domain theories exists for many machine learning applications. In both, feed forward and recurrent neural networks, algortihms for encoding prior knwoledge has been constructed. We propose a heuristic for determining the strength of the prior knowledge (inductive bias) for recurrent neural networks encoded with a DFA as initial domain knowledge. Our heuristic uses gradient information in weight space in the direction of the prior knowledge to enhance performance. Tests on known benchmark problems demonstrate that our heuristic reduces training time, on average, by 30% compared to a random choice of the strength of the inductive bias. It also achieves, on average, near perfect generalization for that specific choice of the inductive bias.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Abu-Mostafa, Y. S.: Learning from Hints in Neural Networks. Journal of Complexity 6 (1990) 192
Fu, L.: Learning Capacity and Sample Complexity on Expert Networks. IEEE Transactions on Neural Networks 7 no. 6 (1996) 1517–1520
Omlin, C. W., Giles, C.: Rule revision with recurrent neural networks. IEEE Transactions on Knowledge and Data Engineering 8 no. 1 (1996) 183–188
Geman, S., Bienenstock, E., Doursat, R.: Neural Networks and the Bias/Variance Dilemma. Neural Computation 4 (1992) 1–58
Snyders, S., Omlin, C. W.: What Inductive Bias Gives Good Neural Network Training Performace? Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Nerworks 3 (2000) 445–450
Tomita, M.: Dynamic construction of finite-state automata from examples, using hill-climbing. Proceedings of the Fourth Annual Cognitive Science Conference (1982) 105–108
Williams, R., Zipser, D.: A learning algorithm for continually running fully recurrent neural networks. Neural Computation 1 (1989) 270–280
Towell, G., Shavlik, J.: Knowledge-based artificial neural networks. Artificial Intelligence 70 no. 1-2 (1994) 119–165
Omlin, C. W., Giles, C.: Constructing Deterministic Finite-State Automata in Recurrent Neural Networks. Journal of the ACM 43 no. 6 (1996) 937–972
Rumelhart, D., Hinton, G., Williams, R.: Learning internal representations by error propagation. Parallel Distributed Processing chp. 8 (1986)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2001 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Snyders, S., Omlin, C.W. (2001). Inductive Bias in Recurrent Neural Networks. In: Mira, J., Prieto, A. (eds) Connectionist Models of Neurons, Learning Processes, and Artificial Intelligence. IWANN 2001. Lecture Notes in Computer Science, vol 2084. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-45720-8_39
Download citation
DOI: https://doi.org/10.1007/3-540-45720-8_39
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-42235-8
Online ISBN: 978-3-540-45720-6
eBook Packages: Springer Book Archive