Study of Two Error Functions to Approximate the Neyman–Pearson Detector Using Supervised Learning Machines

A study of the possibility of approximating the Neyman-Pearson detector using supervised learning machines is presented. Two error functions are considered for training: the sum-of-squares error and the Minkowski error with R = 1. The study is based on the calculation of the function the learning machine approximates to during training, and the application of a sufficient condition previously formulated. Some experiments about signal detection using neural networks are also presented to test the validity of the study. Theoretical and experimental results demonstrate, on one hand, that only the sum-of-squares error is suitable to approximate the Neyman-Pearson detector and, on the other hand, that the Minkowski error with R = 1 is suitable to approximate the minimum probability of error classifier.

[1]  Diego Andina,et al.  A comparison of criterion functions for a neural network applied to binary detection , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[2]  Diego Andina,et al.  Performance improvements for a neural network detector , 1995, Proceedings of ICNN'95 - International Conference on Neural Networks.

[3]  S. Brendle,et al.  Calculus of Variations , 1927, Nature.

[4]  H. Vincent Poor,et al.  Neyman-pearson detection of gauss-Markov signals in noise: closed-form error exponentand properties , 2005, IEEE Transactions on Information Theory.

[5]  Marco Saerens,et al.  Any reasonable cost function can be used for a posteriori probability approximation , 2002, IEEE Trans. Neural Networks.

[6]  Viswanath Ramamurti,et al.  Neural networks for signal detection in non-Gaussian noise , 1997, IEEE Trans. Signal Process..

[7]  Manuel Rosa-Zurera,et al.  NN-Based Detector for Known Targets in Coherent Weibull Clutter , 2006, IDEAL.

[8]  Richard G. Baraniuk,et al.  Controlling False Alarms With Support Vector Machines , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.

[9]  Andrei B. Utkin,et al.  Development of neural network committee machines for automatic forest fire detection using lidar , 2004, Pattern Recognit..

[10]  S. Dudoit,et al.  Comparison of Discrimination Methods for the Classification of Tumors Using Gene Expression Data , 2002 .

[11]  Harry L. Van Trees,et al.  Detection, Estimation, and Modulation Theory: Radar-Sonar Signal Processing and Gaussian Signals in Noise , 1992 .

[12]  Manuel Rosa-Zurera,et al.  Low Complexity MLP-Based Radar Detector: Influence of the Training Algorithm and the MLP Size , 2007, IWANN.

[13]  M. J. D. Powell,et al.  Restart procedures for the conjugate gradient method , 1977, Math. Program..

[14]  H. K. Garg,et al.  Maximum-likelihood detection in DWT domain image watermarking using Laplacian modeling , 2005, IEEE Signal Processing Letters.

[15]  Francisco López-Ferreras,et al.  Combining MLPs and RBFNNs to Detect Signals With Unknown Parameters , 2009, IEEE Transactions on Instrumentation and Measurement.

[16]  David Casasent,et al.  Radial basis function neural networks for nonlinear Fisher discrimination and Neyman-Pearson classification , 2003, Neural Networks.

[17]  Pauli Kuosmanen,et al.  Neural detectors with variable threshold , 1999, ISCAS'99. Proceedings of the 1999 IEEE International Symposium on Circuits and Systems VLSI (Cat. No.99CH36349).

[18]  O.K. Ersoy,et al.  Neural network learning of low-probability events , 1996, IEEE Transactions on Aerospace and Electronic Systems.

[19]  Will Perkins.  ON THE STRONG LAW OF LARGE NUMBERS , 2004 .

[20]  Teddy Furon,et al.  Some theoretical aspects of watermarking detection , 2006, Electronic Imaging.

[21]  E. S. Pearson,et al.  On the Problem of the Most Efficient Tests of Statistical Hypotheses , 1933 .

[22]  H. Vincent Poor,et al.  Neyman-Pearson Detection of Gauss-Markov Signals in Noise: Closed-Form Error Exponent and Properties , 2005, ISIT.

[23]  Bruce W. Suter,et al.  The multilayer perceptron as an approximation to a Bayes optimal discriminant function , 1990, IEEE Trans. Neural Networks.

[24]  David Casasent,et al.  New training strategies for RBF neural networks for X-ray agricultural product inspection , 2003, Pattern Recognit..

[25]  Diego Andina,et al.  Performance Analysis of Neural Network Detectors by Importance Sampling Techniques , 2004, Neural Processing Letters.

[26]  P. P. Gandhi,et al.  Neural detectors for signals in non-Gaussian noise , 1993, 1993 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[27]  M.H.J. Bollen,et al.  A statistical-based sequential method for fast online detection of fault-induced voltage dips , 2004, IEEE Transactions on Power Delivery.

[28]  Paola Sebastiani,et al.  Statistical Challenges in Functional Genomics , 2003 .

[29]  Manuel Rosa-Zurera,et al.  Neural Network Detectors for Composite Hypothesis Tests , 2006, IDEAL.

[30]  Robert D. Nowak,et al.  A Neyman-Pearson approach to statistical learning , 2005, IEEE Transactions on Information Theory.

[31]  Venugopal V. Veeravalli,et al.  Decentralized detection in sensor networks , 2003, IEEE Trans. Signal Process..

[32]  R. Weinstock.  Calculus of Variations: with Applications to Physics and Engineering , 1952 .

[33]  Manuel Rosa-Zurera,et al.  MLP and RBFN for Detecting White Gaussian Signals in White Gaussian Interference , 2003, IWANN.

[34]  E. S. Pearson,et al.  On the Problem of the Most Efficient Tests of Statistical Hypotheses , 1933 .

[35]  S. Haykin,et al.  Signal detection in a nonstationary environment reformulated as an adaptive pattern classification problem , 1998, Proc. IEEE.

[36]  Heekuck Oh,et al.  Neural Networks for Pattern Recognition , 1993, Adv. Comput..

[37]  F. Lopez-Ferreras,et al.  Sufficient condition for an adaptive system to approximate the neyman-pearson detector , 2005, IEEE/SP 13th Workshop on Statistical Signal Processing, 2005.

[38]  A. Gualtierotti.  H. L. Van Trees, Detection, Estimation, and Modulation Theory, , 1976 .