Proceedings Article10.1109/ICASSP.1993.319289
Feature extraction based on minimum classification error/generalized probabilistic descent method
A. Biem,Shigeru Katagiri +1 more
- 27 Apr 1993
- Vol. 2, pp 275-278
62
TL;DR: Although the proposed discriminative feature extraction approach is a direct and simple extension of MCE/GPD, it is a significant departure from conventional approaches, providing a comprehensive basis for the entire system design.
read more
Abstract: A novel approach to pattern recognition which comprehensively optimizes both a feature extraction process and a classification process is introduced. Assuming that the best features for recognition are the ones that yield the lowest classification error rate over unknown data, an overall recognizer, consisting of a feature extractor module and a classifier module, is trained using the minimum classification error (MCE)/generalized probabilistic descent (GPD) method. Although the proposed discriminative feature extraction approach is a direct and simple extension of MCE/GPD, it is a significant departure from conventional approaches, providing a comprehensive basis for the entire system design. Experimental results are presented for the simple example of optimally designing a cepstrum representation for vowel recognition. The results clearly demonstrate the effectiveness of the proposed method. >
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Bearing fault diagnosis based on wavelet transform and fuzzy inference
Xinsheng Lou,Kenneth A. Loparo +1 more
TL;DR: In this article, a new scheme for the diagnosis of localised defects in ball bearings based on the wavelet transform and neuro-fuzzy classification was proposed. But this scheme was only applied to a single motor-driven experimental system, and the results demonstrate that the method can reliably separate different fault conditions under the presence of load variations.
708
Multiple Hypotheses Image Segmentation and Classification With Application to Dietary Assessment
TL;DR: A method for dietary assessment to automatically identify and locate food in a variety of images captured during controlled and natural eating events is proposed and improved accuracy of segmenting food images with classifier feedback is shown.
High performance Chinese OCR based on Gabor features, discriminative feature extraction and model training
Qiang Huo,Yong Ge,Zhi-Dan Feng +2 more
- 07 May 2001
TL;DR: Three key techniques contributing to the high recognition accuracy are highlighted, namely theuse of Gabor features, the use of discriminative feature extraction, and theUse of minimum classification error as a criterion for model training.
Minimum classification error training for online handwriting recognition
TL;DR: An HMM-based, character and word-level MCE training aimed at minimizing the character or word error rate while enabling flexibility in writing style through the use of multiple allographs per character is described.
70
Neural pattern identification of railroad wheel-bearing faults from audible acoustic signals: comparison of FFT, CWT, and DWT features
Howard C. Choe,Yulun Wan,Andrew K. Chan +2 more
- 03 Apr 1997
TL;DR: This paper presents a novel method to detect, recognize, and classify a variety of railroad wheel-bearing defects using audible acoustic signals at several different train speeds.
69
References
On the use of bandpass liftering in speech recognition
Biing-Hwang Juang,Lawrence R. Rabiner,Jay G. Wilpon +2 more
- 07 Apr 1986
TL;DR: It is found that measurements of speech spectral envelopes are prone to statistical variations due to window position fluctuations, excitation interference, measurement noise, etc. and may possess spurious characteristics because of analysis model constraints and that a statistical model can be established to predict the variances of the cepstral coefficient measurements.
217
New discriminative training algorithms based on the generalized probabilistic descent method
Shigeru Katagiri,C.-H. Lee,Biing-Hwang Juang +2 more
- 30 Sep 1991
TL;DR: A family of new discriminative training algorithms can be rigorously formulated for various kinds of classifier frameworks, including the popular dynamic time warping (DTW) and hidden Markov model (HMM).
164
A weighted cepstral distance measure for speech recognition
Y. Tohkura
- 01 Apr 1986
TL;DR: The experimental results show that the weighted cepstral distance measure works substantially better than both the Euclidean cepStral distance and the log likelihood ratio distance measures across two different data bases, namely a 10 digits and a 129 airline vocabulary words.
135
Prototype-based discriminative training for various speech units
Erik McDermott,Shigeru Katagiri +1 more
- 23 Mar 1992
TL;DR: The authors extend LVQ into a prototype-based classifier appropriate for the classification of various long speech units, and their results reveal clear gains in performance as a result of using PBMEC.
36
Discriminative template training for dynamic programming speech recognition
Pao-Chung Chang,Biing-Hwang Juang +1 more
- 23 Mar 1992
TL;DR: A newly proposed minimum recognition error formulation and a generalized probabilistic descent (GPD) algorithm are analyzed and used to accomplish discriminative training of a conventional dynamic programming based speech recognizer.
27