Showing papers on "Word error rate published in 1972"

Considerations of sample and feature size

Abstract: In many practical pattern-classification problems the underlying probability distributions are not completely known. Consequently, the classification logic must be determined on the basis of vector samples gathered for each class. Although it is common knowledge that the error rate on the design set is a biased estimate of the true error rate of the classifier, the amount of bias as a function of sample size per class and feature size has been an open question. In this paper, the design-set error rate for a two-class problem with multivariate normal distributions is derived as a function of the sample size per class (N) and dimensionality (L) . The design-set error rate is compared to both the corresponding Bayes error rate and the test-set error rate. It is demonstrated that the design-set error rate is an extremely biased estimate of either the Bayes or test-set error rate if the ratio of samples per class to dimensions (N/L) is less than three. Also the variance of the design-set error rate is approximated by a function that is bounded by 1/8N .

Orthographic Distinctiveness as a Variable in Word Recognition.

Abstract: Of low-frequency words scaled for orthographic distinctiveness, the structurally interesting or unusual (high distinctiveness) words were recognized better, and were chosen less often when used as distracters, than words rated not distinctive in orthography. Confidence judgments in correct recognition responses were significantly affected by the presence of a distinctive test distracter. Orthographic distinctiveness may be encoded as part of the memory for a word.

Application of optimum error-reject functions (Corresp.)

Abstract: In an optimum pattern-recognition system the error rate is determined by the reject function. This correspondence describes how this property may be exploited to provide quantitative tests of model validity using unclassified test samples. These tests are basically goodness-of-fit tests for a function of the observations. One of these tests is shown to provide an improved estimate of error in Monte Carlo studies of complex systems. Results are given for normal distributions when parameters are estimated. In this case error estimates obtained from the empirical reject rate underestimate the actual error and performance depends on the ratio of design samples to dimension.

Digital servo motor control with an error rate derivative circuit

Abstract: A serial digital word circuit utilizing basic summing, multiplying, sign detection and overflow detection circuits for providing an error rate derivative circuit. The error rate derivative is accomplished by comparing a present digital word time serial digital word feedback signal with a previous digital word time serial digital word feedback signal and adding this to a comparison of the feedback signal and a command signal. As the feedback signal indicates a controlled condition is approaching the commanded condition, the sum of the added digital words reduces in magnitude and thus the effect of the control signal is more rapidly reduced as the condition reaches a desired value.

Rapid processing of multispectral scanner data using linear techniques.

Abstract: Tests have been made to compare linear and quadratic techniques for processing multispectral scanner data. The tests have been limited to a few selected sets of agricultural data. Two aspects of processing were studied. The first, the selection of a subset of channels to be used in the decision function, was found to be faster by a factor of 50 when a linear method was used. Second, in recognition processing, our linear decision rule produced a lower error rate and utilized a larger number of channels for equal processing times. Nevertheless, when the criterion is lowest possible error rate, irregardless of processing time, the quadratic rule is preferable.

Concatenated coding for low date rate space communications.

Abstract: In deep space communications with distant planets, the data rate as well as the operating SNR may be very low. To maintain the error rate also at a very low level, it is necessary to use a sophisticated coding system (longer code) without excessive decoding complexity. The concatenated coding has been shown to meet such requirements in that the error rate decreases exponentially with the overall length of the code while the decoder complexity increases only algebraically. Three methods of concatenating an inner code with an outer code are considered. Performance comparison of the three concatenated codes is made.