Abstract
To train Neural Networks (NNs) in a supervised way, estimations of an objective function must be carried out. The value of this function decreases as the training progresses and so, the number of test observations necessary for an accurate estimation has to be increased. Consequently, the training computational cost is unaffordable for very low objective function value estimations, and the use of Importance Sampling (IS) techniques becomes convenient. The study of three different objective functions is considered, which implies the proposal of estimators of the objective function using IS techniques: the Mean-Square error, the Cross Entropy error and the Misclassification error criteria. The values of these functions are estimated by IS techniques, and the results are used to train NNs by the application of Genetic Algorithms. Results for a binary detection in Gaussian noise are provided. These results show the evolution of the parameters during the training and the performances of the proposed detectors in terms of error probability and Receiver Operating Characteristics curves. At the end of the study, the obtained results justify the convenience of using IS in the training.
Similar content being viewed by others
References
Atanassov E, Dimov IT (2007) What Monte Carlo models can do and cannot do efficiently?. Appl Math Model 32: 1477–1500
Borcherds PH (2000) Importance sampling: an illustrative introduction. Eur J Phys 21: 405–411
Denny M (2001) Introduction to importance sampling in rare-event simulations. Eur J Phys 22: 403–411
Andrieu C, Freitas N, Doucet A, Jordan MI (2003) An introduction to MCMC for machine learning. Mach Learn 50: 5–43
Bengio Y, Senécal JS (2003) Quick training of probabilistic neural nets by importance sampling. In: Proceedings of artificial intelligence statistics 2003 (AIStats 2003)
Bengio Y, Senécal JS (2008) Adaptive importance sampling to accelerate training of a neural probabilistic language model. IEEE Trans Neural Netw 19: 713–722
Yuan C, Druzdzel MJ (2006) Importance sampling algorithms for Bayesian networks: principles and performance. Math Comput Model 43: 1189–1207
Blumensath T, Davies M (2005) A fast importance sampling algorithm for unsupervised learning of over-complete dictionaries. In: Proceedings of the IEEE international conference on acoustics, speech, and signal processing, 2005 (ICASSP ‘05), March 18–23, vol 5, pp 213–216
Uchibe E, Doya K (2004) Competitive-cooperative-concurrent reinforcement learning with importance sampling. In: Proceedings of the 8th international conference on simulation of adaptive behavior: from animals and animats, July 13–17, pp 287–296
Chen JC, Lu D, Sadowsky JS, Yao K (1993) On importance sampling in digital communications. Part I: fundamentals. IEEE J Sel Areas Commun 11: 289–299
Wolfe RJ, Jeruchim MC, Hahn PM (1990) On optimum and suboptimum biasing procedures for importance sampling in communication simulation. IEEE Trans Commun 38: 639–647
Al-Qaq WA, Devetsikiotis M, Townsend JK (1995) Stochastic gradient optimization of importance sampling for the efficient simulation of digital communication systems. IEEE Trans Commun 43: 2975–2985
Smith PJ, Shafi M, Gao H (1997) Quick simulation: a review of importance sampling techniques in communications systems. IEEE J Sel Areas Commun 15: 597–613
Al-Qaq WA, Townsend JK (1997) A stochastic importance sampling methodology for the efficient simulation of adaptive systems in frequency nonselective Rayleigh fading channels. IEEE J Sel Areas Commun 15: 614–625
Orsak GC, Aazhang B (1991) Constrained solutions in importance sampling via robust statistics. IEEE Trans Inf Theory 37: 307–316
Sanz-González JL, Andina D (1999) Performance analysis of neural network detectors by importance sampling techniques. Neural Process Lett 9: 257–269
Gerlach K (1999) New results in importance sampling. IEEE Trans Aerosp Electron Syst 35: 917–925
Orsak GC (1993) A note on estimating false alarm rates via importance sampling. IEEE Trans Commun 41: 1275–1277
Hoogerheide LF, Kaashoek JF, van Dijk HK (2007) On the shape of posterior densities and credible sets in instrumental variable regression models with reduced rank: an application of flexible sampling methods using neural networks. J Econom 139: 154–180
Hoogerheide LF, van Dijk HK (2010) Bayesian forecasting of value at risk and expected shortfall using adaptive importance sampling. Int J Forecast 26(2): 231–247
Gandhi PP, Ramamurti V (1997) Neural networks for signal detection in non-Gaussian noise. IEEE Trans Signal Process 45: 2846–2851
Andina D, Sanz-González JL (1996) Comparison of a neural network detector vs. Neyman–Pearson optimal detector. In: Proceedings of the IEEE international conference on acoustics, speech, and signal processing (ICASSP’96), May 7–10, vol 6, pp 3573–3576
Jarabo-Amores MP, Rosa-Zurera M, Gil-Pita R, López-Ferreras F (2009) Study of two error functions to approximate the Neyman–Pearson detector using supervised learning machines. IEEE Trans Signal Process 57: 4175–4181
Sanz-González JL, Andina D (2001) Importance sampling techniques in neural detector training. Lect Notes Comput Sci 2167: 431–436
Sanz-González JL, Andina D, Seijas J (2002) Importance sampling and mean-square error in neural detector training. Neural Process Lett 16: 259–276
Rosa-Zurera M, Jarabo-Amores P, López-Ferreras F, Sanz-González JL (2005) Comparative analysis of importance sampling techniques to estimate error functions for training neural networks. In: Proceedings of the IEEE 13th workshop on statistical signal processing, July 17–20, pp 121–125
Hachiya H, Akiyama T, Sugiayma S, Peters J (2009) Adaptive importance sampling for value function approximation in off-policy reinforcement learning. Neural Netw 22: 1399–1410
Poor HV (1994) An introduction to signal detection and estimation, 2nd edn. Springer, Berlin
Hampshire JB, Pearlmutter B (1990) Equivalence proofs for multilayer perceptron classifiers and the Bayesian discriminant function. In: Proceedings of the 1990 Connectionist Models Summer School, pp 159–172
Van Trees HL (2003) Detection, estimation and modulation theory. Part I. Wiley, New York
Haykin S (2008) Neural networks and learning machines, 3rd edn. Prentice-Hall, London
Seiffer U (2001) Multiple layer perceptron training using genetic algorithms. In: Proceedings of the 2001 European symposium on artificial neural networks, pp 159–164
Montana DJ, Davis L (1989) Training feedforward neural networks using genetic algorithms. In: Proceedings of 11th international joing conference on artificial intelligence, pp 762–767
Nguyen D, Widrow B (1990) Improving the learning speed of 2-layer neural networks by choosing initial values of the adaptive weights. In: Proceedings of the 1990 international joint conference on neural networks (IJCN 1990), June 17–21, vol 3, pp 21–26
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Vicen-Bueno, R., Jarabo-Amores, M.P., Rosa-Zurera, M. et al. Importance Sampling for Objective Function Estimations in Neural Detector Training Driven by Genetic Algorithms. Neural Process Lett 32, 249–268 (2010). https://doi.org/10.1007/s11063-010-9155-8
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11063-010-9155-8