一種概率映射網(wǎng)絡(luò)的EM訓(xùn)練算法
AN EFFICIENT EM TRAINING ALGORITHM FOR PROBABILITY MAPPING NETWORKS
-
摘要: 文中提出一種概率映射網(wǎng)絡(luò)(PMN)的EM(Expectation Maximization)訓(xùn)練算法。PMN為一個(gè)四層前饋網(wǎng)。它構(gòu)成一個(gè)貝葉斯分類器，實(shí)現(xiàn)多類分類的貝葉斯判別，把輸入的樣本模式經(jīng)網(wǎng)絡(luò)變換為輸出的分類判決，其網(wǎng)絡(luò)節(jié)點(diǎn)對(duì)應(yīng)于貝葉斯后驗(yàn)概率公式的各個(gè)變量。 此PMN用高斯核函數(shù)作為密度函數(shù)，網(wǎng)絡(luò)參數(shù)訓(xùn)練由EM算法實(shí)現(xiàn)，其學(xué)習(xí)方式為類間的監(jiān)督學(xué)習(xí)和類內(nèi)的非監(jiān)督學(xué)習(xí)。最后的實(shí)驗(yàn)表明此網(wǎng)絡(luò)及其學(xué)習(xí)算法在分類應(yīng)用中的有效性。Abstract: An Expectation-Maximization(EM) training algorithm for estimating the parameters of a special Probability Mapping Network (PMN) structure which forms a multicatolog Bayes classifier is proposed in this paper. The structure of PMN is a four-layer Feedforward Neural Networks(FNN), where the Gaussian probability density function is realized as an internal node. In this way, the EM algorithm is extended to deal with supervised learning of a multicatolog of the neural network Gaussian classifier. The computational efficiency and the numerical stability of the training algorithm benefit from the well-established EM framework. The effectiveness of the proposed network architecture and its EM training algorithm are assessed by conducting two experiments.
-
Lee A S, IGl R M. A Gaussian potential function network with hierarchically self-organizing learning. Neural Networks, 1991 4(1): 207-224.[2] Speht P F. Probabilistic neural networks. Neural Networks, 1990, 3(1): 109-118.[2]Ma Sheng, Ji Chuanyi, Farmer J. An efficient EM-based training algorithm for feedforward neural networks. Neural Networks, 1997, 10(2), 243-256.[4] Streit R L, Luginbahl T E Maximum likelihood training of probabilistic neural networks. IEEE Trans. on NN, 1994, NN-5(5): 764-783.[3]Traven G C. A neural network approach to statistical pattern classfication by semi-parametric estimation of probability density function. IEEE Trans. on NN, 1991, NN-2(3): 366-377.[4]Dempster A P, Laird N M, Rubin D R. Maximum likelihood from incomplete date via the EM algorithm. J. Royal Statiscal Sac.,Ser. B, 1977; 39(1): 1-38.[7] Liporace L A. Maximum likelilood estimation for multivariate observations of Markov sources. IEEE Trans. on IT, 1982, IT-28(5): 729-734.[8] Wu J, Chan C. Isolated word recogontion by neural network models with crosscorrelation coef- ficient for speech recognition. IEEE Trans. on PAMI, 1993, PAMI-15(11): 1174-1185. -
計(jì)量
- 文章訪問數(shù): 732
- HTML全文瀏覽量: 136
- PDF下載量: 574
- 被引次數(shù): 0
下載:
