Title :
Minimum Probability of Error Signal Representation
Author :
Silva, Jorge ; Narayanan, Shrikanth
Author_Institution :
Univ. of Southern California, Los Angeles
Abstract :
The problem of minimum probability of error signal representation (MPE-SR) considering issues of finite training data is revisited and extended in this paper. Results are presented that justify addressing this problem as a complexity-regularized optimization criterion, reflecting the well-known tradeoff between signal representation quality and learning complexity. A rate-distortion type of formulation is proposed to address this optimization problem by finding a sequence of signal representations achieving optimal complexity-fidelity operational points. Finally under specific assumptions, it is shown that the MPE-SR reduces to a version of Fisher linear discriminant analysis.
Keywords :
optimisation; probability; signal representation; Fisher linear discriminant analysis; complexity-regularized optimization criterion; error signal representation; finite training data; learning complexity; minimum probability; signal representation quality; Data engineering; Degradation; Laboratories; Linear discriminant analysis; Radio access networks; Random variables; Rate-distortion; Signal analysis; Signal representations; Viterbi algorithm;
Conference_Titel :
Machine Learning for Signal Processing, 2007 IEEE Workshop on
Conference_Location :
Thessaloniki
Print_ISBN :
978-1-4244-1566-3
Electronic_ISBN :
1551-2541
DOI :
10.1109/MLSP.2007.4414331
