scispace - formally typeset
Search or ask a question

Showing papers on "Feature (machine learning) published in 1975"


01 Dec 1975
TL;DR: MSYS is a system for reasoning with uncertain information and inexact rules of inference that contributes to the interpretation of visual features in scene analysis by repeatedly applying rules until a consistent set of likelihood values is attained.
Abstract: : MSYS is a system for reasoning with uncertain information and inexact rules of inference. Its major application, to date, has been to the interpretation of visual features (such as regions) in scene analysis. In this application, features are assigned sets of possible interpretations with associated likelihoods based on local attributes (e.g., color, size, and shape). Interpretations are related by rules of inference that adjust the likelihoods up or down in accordance with the interpretation likelihoods of related features. An asynchronous relaxation process repeatedly applies the rules until a consistent set of likelihood values is attained. At this point, several alternative interpretations still exist for each feature. One feature is chosen and the most likely of its alternatives is assumed. The rules are then used in this more precise context to determine likelihoods for the interpretations of remaining features by a further round of relaxation. The selection and relaxation steps are repeated until all features have been interpreted.

88 citations


Journal ArticleDOI
TL;DR: Although all three stages of the pattern recognition system play an essential role in the process of classifying patterns by machine, the quality of the system's performance depends chiefly on the feature selector, which has a beneficial effect on the progress in the theory of pattern recognition.
Abstract: In the 15 years of its existence pattern recognition has made considerable progress on both the theoretical and practical fronts. Starting from the original application of pattern recognition techniques to the problem of character recognition at the time when pattern recognition was conceived these techniques have now penetrated such diverse areas of science as medical diagnosis, remote sensing, finger prints and speech recognition, image classification, etc.* This wide applicability derives from the inherent generality of pattern recognition, which is a direct consequence of the adopted threestage concept of pattern recognition process. According to this concept the process of pattern recognition is viewed as a sequence of three independent functions--representation, feature selection and classification (Fig. 1). Among these functions only the representation stage, which transforms the input patterns into a form suitable for computer processing, is problemdependent. Both the feature selector, the function of which is to reduce the dimensionality of the representation vector, and the classifier, which carries out the actual decision process, work with a vector of measurements which can be considered as an abstract pattern. As a result, the feature selection and classification stages can be implemented using mathematical methods irrespective of the original application. Naturally, this has had a beneficial effect on the progress in the theory of pattern recognition. Although all three stages of the pattern recognition system play an essential role in the process of classifying patterns by machine, the quality of the system's performance depends chiefly on the feature selector. The reasons

80 citations


Journal ArticleDOI
TL;DR: The methods developed in this correspondence represent an approach to the problem of handling error-corrupted syntactic pattern strings, an area generally neglected in the numerous techniques for linguistic pattern description and recognition which have been reported.
Abstract: The methods developed in this correspondence represent an approach to the problem of handling error-corrupted syntactic pattern strings, an area generally neglected in the numerous techniques for linguistic pattern description and recognition which have been reported. The basic approach consists of applying error transformations to the productions of context-free grammars in order to generate new grammars (also context-free) capable of describing not only the original error-free patterns, but also patterns containing specific types of errors such as deleted, added, and interchanged symbols which arise often in the pattern-scanning process. Theoretical developments are illustrated in the framework of a syntactic recognition system for chromosome structures.

40 citations



Patent
Hiroaki Sakoe1
30 May 1975
TL;DR: In this article, a dynamic matching unit finds out a maximum of degrees of coincidence between an input character and line patterns derived from each of the standard patterns by varying the magnitudes of the vectors between predetermined minima and maxima.
Abstract: A character recognition device comprises a memory for memorizing standard patterns, each given by fundamental vectors. A dynamic matching unit finds out a maximum of degrees of coincidence between an input character and line patterns derived from each of the standard patterns by varying the magnitudes of the vectors between predetermined minima and maxima and derives similarity measures defined by the maximum degrees of coincidence between the input character and the line patterns derived from the respective standard patterns and specific line patterns for which the maximum degrees of coincidence are found. A decision unit compares the similarity measures with one another, judges whether or not the input character has a feature predetermined for each of the standard patterns, and delivers a result of the recognition in response to results of the comparison and judgment.

29 citations


Journal ArticleDOI
TL;DR: In this paper, a mathematical formulation for each of several zero-crossing feature extraction techniques is derived and related (where possible) to each of the other zero-Crossing methods.
Abstract: Zero-crossing analysis techniques have long been applied to speech analysis, to automatic speech recognition, and to many other signal-processing and pattern-recognition tasks. In this paper, a mathematical formulation for each of several zero-crossing feature extraction techniques is derived and related (where possible) to each of the other zero-crossing methods. Based upon this mathematical formulation, a physical interpretation of each analysis technique is effected, as is a discussion of the properties of each method. It is shown that four of these methods are a description of a short-time waveform in which essentially the same information is preserved. Each turns out to be a particular normalization of a count of zero-crossing intervals method. The effects of the various forms of normalization are discussed. A fifth method is shown to be a different type of measure; one which preserves information concerning the duration of zero-crossing intervals rather than their absolute number. Although reference is made as to how each of the zero-crossing methods has been applied to automatic speech recognition, an attempt is made to enumerate general characteristics of each of the techniques so as to make the mathematical analysis generally applicable.

25 citations


Proceedings Article
03 Sep 1975
TL;DR: The theory is used to obtain the a priori probabilities that are necessary in the application of stochastic languages to pattern recognition and A.I. theory to make critical decisions.
Abstract: Recent results in induction theory are reviewed that demonstrate the general adequacy of the induction system of Solomoncff and Willis. Several problems in pattern recognition and A.I. are investigated through these methods. The theory is used to obtain the a priori probabilities that are necessary in the application cf stochastic languages to pattern recognition. A simple, quantitative solution is presented for part of Winston's problem of learning structural descriptions from exandples. In contrast to work in non-probabilistic prediction, the present methods give probability values that can be used with decision. theory to make critical decisions.

17 citations


Journal ArticleDOI
M. Sambur1
TL;DR: The effectiveness of a set of speaker recognition features is usually characterized in terms of the ratio of the interspeaker variability of the feature to its intraspeaker variability (F‐ratio), but by an appropriate eigenvector analysis, aSet of orthogonal parameters can be obtained that is essentially constant across an utterance for a given speaker.
Abstract: The effectiveness of a set of speaker recognition features is usually characterized in terms of the ratio of the interspeaker variability of the feature to its intraspeaker variability (F‐ratio). A recent experiment in speech synthesis [M.R. Sambur, “An Efficient LPC Vocoder,” Bell Syst. Tech. J. (to be published)] has shown that by an appropriate eigenvector analysis, a set of orthogonal parameters can be obtained that is essentially constant across an utterance for a given speaker (i.e., zero intraspeaker variability). If the same eigenvector analysis is applied to the same utterance spoken by another speaker, the resulting values of the orthogonal parameters are, however, different. These orthogonal parameters were therefore examined for their ability to differentiate different speakers. They were formally tested in a speaker recognition experiment involving 21 speakers. The speech data consisted of six repetitions of the same sentence spoken by each speaker on six separate occasions. The identificatio...

14 citations


Journal ArticleDOI
TL;DR: The use of multiple form dimensions in pattern classification was studied with adults and children in grades 2 and 5, and multiple feature use in classification was evidenced at all age levels.
Abstract: The use of multiple form dimensions in pattern classification was studied with adults and children in grades 2 and 5. Each subject sorted 30 8-sided random polygons first into 2, then into 3, and finally into 4 groups and repeated the procedure 1 week later. A series of discriminant analyses, using 9 physical form characteristics as predictors, was used to answer several developmental questions. Reliability of classification, number and saliency of features selected, and accuracy with which they were used all implied continuous development of perceptual skills. Multiple feature use in classification was evidenced at all age levels.

10 citations


Book ChapterDOI
01 Jan 1975
TL;DR: This chapter focuses on the nature of the primary recognition process and the acoustic features of speech stimuli that are utilized in the speech perception process.
Abstract: Publisher Summary According to an information-processing analysis, understanding the spoken word involves a series of psychological processes between first detecting the features of the acoustic wave form and finally manipulating the ideas in the speaker's message. This chapter focuses on the nature of the primary recognition process and the acoustic features of speech stimuli that are utilized in the speech perception process. An incoming speech signal is initially stored as a brief preperceptual auditory image. The process of perception begins with the extraction of information from that image. Then, on the basis of that information, a decision is made, resulting in the recognition or synthesis of the speech pattern as a unique unit of roughly syllabic size. Combining these units into words and sentences is then accomplished by later stages of processing. Any phoneme is marked not by a single unique acoustic characteristic but rather by a set of characteristics, some of which may occur in other phonemes. Such characteristics are called acoustic features if they are employed in the recognition process. An acoustic characteristic qualifies as an acoustic feature when the presence or absence of that characteristic is critical to the recognition process.

9 citations


01 Jan 1975
TL;DR: A distributed processing network for pattern recognition is developed for use as a syntax system for speech understanding and an introduction to the problems of representation and recognition of structured patterns is presented.
Abstract: : This collection of four papers presents many of the major findings in the authors research related to the representation, recognition, and learning of structured patterns. The first paper presents an introduction to the problems of representation and recognition of structured patterns and suggests several directions for further research. One of these suggestions concerns a distributed processing network for pattern recognition. In the second paper (co-authored by D. J. Mostow), such a network is developed for use as a syntax system for speech understanding. The last two papers are concerned specifically with the learning problems encountered in the framework of structured patterns.

Journal ArticleDOI
TL;DR: Evidence is presented to suggest that the feature analytic process is insufficient to account for the phenomenon of pattern recognition, and it is suggested that two processes are necessary prior to feature analysis.

Journal ArticleDOI
TL;DR: Test implementation of this scheme using the remotely sensed agricultural data of the Purdue laboratory for agricultural remote sensing in a simulated unsupervised mode, has brought out the efficacy of this integrated system of feature selection and learning.
Abstract: Here the twin problems of feature selection and learning are tackled simultaneously to obtain a unified approach to the problem of pattern recognition in an unsupervised environment. This is achieved by combining a feature selection scheme based on the stochastic learning automata model with an unsupervised learning scheme such as learning with a probabilistic teacher. Test implementation of this scheme using the remotely sensed agricultural data of the Purdue laboratory for agricultural remote sensing (LARS) in a simulated unsupervised mode, has brought out the efficacy of this integrated system of feature selection and learning.

Journal ArticleDOI
TL;DR: The letter describes a method of weighting features so that the significant parameters of the input patterns are emphasised, which enables dimensionality reduction to take place without loss of information necessary to distinguish between classes.
Abstract: Feature extraction remains a problem in pattern recognition, and the letter describes a method of weighting features so that the significant parameters of the input patterns are emphasised. This enables dimensionality reduction to take place without loss of information necessary to distinguish between classes.

Journal ArticleDOI
TL;DR: A Markov chain feature transition model of capital letter recognition is presented for synthesizing the empirically derived 26 × 26 capital letter confusion matrices to minimize the sum of per cell squared differences between the empirical confusion matrix and the synthesized matrix.

Proceedings Article
K. Shirai1
03 Sep 1975
TL;DR: A feature extraction method for speech waves and an algorithm for sentence recognition are studied, based on an articulatory model constructed from the statistical analysis of X-ray data.
Abstract: A feature extraction method for speech waves and an algorithm for sentence recognition are studied. The feature extraction is based on an articulatory model constructed from the statistical analysis of X-ray data. The model holds implicitly the physiological constraints and made possible to estimate the state of the articulatory mechanism. The estimated articulatory parameters provide a set of good features for the speech recognition. The sentence recognition problem is mathematically formulated as an optimization problem with constraints by introducing sentence structures from the syntactic and semantic considerations. The algorithm presents an optimal solution in the Bayesian sense.

Journal ArticleDOI
TL;DR: By storing training samples, where a sample is a feature vector, the class conditional probability density is estimated and the procedure is applied to surgical patient features, the features corresponding to the kind of operation, diagnosis, patients' age, patient's sex, etc.

Journal ArticleDOI
TL;DR: An experiment is reported in which an interactive research tool—a Digital Pattern Playback (DPP)—is being used to evaluate a spectrum‐matching and dictionary‐search technique for speech recognition.
Abstract: Facilities which make spectrograms immediately available for visual comparison, easy modification of spectral data, and resynthesis of speech have proved to be particularly useful tools in speech research. This paper reports an experiment in which such an interactive research tool—a Digital Pattern Playback (DPP)—is being used to evaluate a spectrum‐matching and dictionary‐search technique for speech recognition. The DPP is a computer‐supported analysis‐synthesis facility which, in the present experiment, displays spectrograms of “unknown” sentences so that an analyst can list the important acoustic features of marked segments of the unknown sentence. Interrogation of a feature‐based dictionary then recovers all items with features which match the unknown segment. If necessary, additional features may be assigned to narrow the search. The reference spectrograms retrieved from the dictionary are compared, one at a time, with the spectrogram of the unknown sentence and the best match is selected for each un...

01 Dec 1975
TL;DR: This work has shown that the availability of stress information permitted more accurate vowel recognition by determining beforehand the reliability of a phoneme's formant structure, word boundary detection, and a partial estimate of a passage's syntactic structure.
Abstract: : Past research in speech recognition has emphasized the need to apply prosodic feature information when making decisions at the phonetic and lexical levels. Of particular interest has been a means by which the stress contour of a continuous speech sentence could be obtained without first performing phonetic recognition. The availability of such stress information permitted: (1) More accurate vowel recognition by determining beforehand the reliability of a phoneme's formant structure, (2) Word boundary detection, and (3) A partial estimate of a passage's syntactic structure.