C. G. Hilborn and D.G. Laintorts, "Unsupervised Learning Minimum Risk Pattern Classification for Dependent Hypotheses and Depedent Measurements", IEEE Trans. on Systems Science and Cybernetics, vol.5, pp. 109-115, April 1969.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of classes, using the full set of past data without preprocessing to perform a recursire computation of membership function which is based on predictive accuracy. The inspiration for PREMOFS comes fi om the Patiition Algorithm, a method originally applied to system identification and control [13, 22, 23, 36]. We have used the Partition Algorithm as the basis for developing the PREMONNS (PREdictive Modular Neural Networks) family of neural classification algorithms [18, 31, 32] these are close relatives of PREMOFS. Both the Partition Algorithm and the original PREMONN [31] are based on a Bwesian ....

C. G. Hilborn and D.G. Laintorts, "Unsupervised Learning Minimum Risk Pattern Classification for Dependent Hypotheses and Depedent Measurements", IEEE Trans. on Systems Science and Cybernetics, vol.5, pp. 109-115, April 1969.

...., Yt. In this paper we do the following. 1. We develop a recursive algorithm to compute the posterior probabil ities. This algorithm combines elements of two algorithms previously reported in the literature: the Backward Fowvard algorithm of Baum [5, 6] and the Partition Algorithm of Lainiotis [13, 17]. 2 2. We prove the convergence of our classification algorRhm: the posterior probability of the best in a precisely defined sense) candidate model tends to one almost surely. 3. We present examples of classification using speech data and phoneme Hidden Markov Models. Hidden Markov Models ....

....about their universal representation and consistent estimation properties (see [14] Our classification algorithm is modelled after the Partition Algorithm of Lainiotis, which has been used for parameter estimation of stochastic control systems. An early version of this algorithm can be found in [13, 17]. This is a very general algorithm, which applies to continuous as well as discrete valued stochastic processes. However, details for the discrete valued case have not been worked out in the literature. Control theoretic applications can be found in [11, 17, 18, 19] and computational issues are ....

[Article contains additional citation context not shown here]

C.G. Hilborn and D.G. Laintorts, "Unsupervised learning minimum risk pattern classification for dependent hypotheses and dependent measure- ments, lEEIF Trans. on Systems Science and Cybernetics, Vol.5, No.2, April 1969, pp. 109-115.

....of such flnite state stochastic automata are considered and an algorithm is developed for estimating their parameters. Also prediction and classiflcation algorithms are developed and applied to modelling of speech data. This work is reported in [Keh91a] and is strongly in uenced by [HL69a] and [HL69b] There is a lot of empirical work done on the question of estimating HMM s. However, the underlying theory is not very well developed. The previously mentioned work by Baum and collaborators develops the theory of a particular maximization technique (namely the BF algorithm) Also the question ....

C. G. Hilborn and D.G.Lainiotis. Unsupervised Learning Minimum Risk Pattern Classiflcation for Dependent Hypotheses and Dependent Measurements. IEEE Trans. PAMI, 2:109-115, 1969.

No context found.

C. G. Hilborn and D.G. Lainiotis. 1969. "Unsupervised Learning Minimum Risk Pattern Classifi- cation for Dependent Hypotheses and Depedent Measurements", IEEE Trans. on Systems Science and Cybernetics, vol.5, pp. 109-115.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC