Browsing by Author "HONG, TP"

Now showing 1 - 4 of 4

AN OPTIMAL PARALLEL PERCEPTRON LEARNING ALGORITHM FOR A LARGE TRAINING SET
(1994-03-01)
HONG, TP
;
TSENG, SS
;
資訊工程學系
;
Department of Computer Science
In [2], a parallel perceptron learning algorithm on the single-channel broadcast communication model was proposed to speed up the learning of weights of perceptrons [3]. The results in [2] showed that given n training examples, the average speedup is 1.48*n0.91/log n by n processors. Here, we explain how the parallelization may be modified so that it is applicable to any number of processors. Both analytical and experimental results show that the average speedup can reach nearly O(r) by r processors if r is much less than n.
ARRANGING APPROPRIATE TRAINING ORDER IN NOISY DOMAINS
(1994)
HSU, YT
;
TSENG, SS
;
HONG, TP
;
資訊工程學系
;
Department of Computer Science
LEARNING CONCEPTS IN PARALLEL BASED UPON THE STRATEGY OF VERSION SPACE
(1994-12-01)
HONG, TP
;
TSENG, SS
;
資訊工程學系
;
Department of Computer Science
In this paper, we have attempted to apply the technique of parallel processing to concept learning. A parallel version-space learning algorithm based upon the principle of divide-and-conquer is proposed. Its time complexity is analyzed to be O(k log(2) n) with n processors, where n is the number of given training instances and k is a coefficient depending on application domains. For a bounded number of processors in the real situations, a modified parallel learning algorithm is then proposed. Experimental results are then performed on a real learning problem, showing our parallel learning algorithm works and being quite consistent with results of theoretic analysis. We have finally concluded that when the number of training instances is large, it is worth learning in parallel because of its faster execution.
PARALLEL PERCEPTRON LEARNING ON A SINGLE-CHANNEL BROADCAST COMMUNICATION MODEL
(1992-02-01)
HONG, TP
;
TSENG, SS
;
資訊科學與工程研究所
;
Institute of Computer Science and Engineering
A parallel perceptron learning algorithm based upon a single-channel broadcast communication model has been proposed here. Since it can process training instances in parallel, instead of one by one in the conventional algorithm, large speedup can be expected. Theoretical analysis shows: with n processors, the average speedup ranges from O(log n) to O(n) under a variety ot assumptions (where n is the number of training instances). Experimental results further show the actual average speedup is approximately being O(n0.91/log n). Extensions to a bounded number of processors and to the backpropagation learning have also been discussed.