| J. Ghosh, L. Deuser, and S. Beck, "A Neural Network based Hybrid System for Detection, Characterization and Classification of Short-duration Oceanic Signals," IEEE Journal of Ocean Engineering, vol. 17, no. 4, pp. 351--363, October 1992. |
....NEW data through the first two trained machines. If the two machines disagree, add this data to the training set for the third machine. b) Train the third machine. 4. vote for committee output. Compare the two outputs : f 1 (x) 0:23 0:25 0:87) f 2 (x) 0:43 0:45 0:52) The entropy fusion [Ghosh et al. 1992] given c classifiers ( independent ) The classier number i has its output j set to y ij . You can normalise the output computing the normalized output z ij = y ij P n k=1 y ik then the entropy of the classifier i is e i = n X k=1 z ik log z ik the fusion of the classifiers is s j = ....
Ghosh, J., Deuser, L., and Beck, S. (1992). A neural network based hybrid system for detection characterisation and classification of short-duration oceanic signals. IEEE J. of ocean engineering, 17(4):351--363.
....is called data set 2 (DS2) Table 1 lists the number of signals per class for the training and test subsets of each data set. Table 2 lists the average number of feature vectors per signal for each signal class. Each feature vector is 8 dimensional and denotes signal energy in 8 frequency bands [6]. The classes 1 4 in data set 1 are obviously not the same as the classes 1 4 in data set 2, but the have been tabulated as if they were, in order to create compact tables. Our investigation demonstrates that MLPs and TDNNs with habituation preprocessing units outperform unhabituated TDNNs of ....
Ghosh, J., Deuser, L., Beck, S. "A Neural Network Based Hybrid System for Detection, Characterization, and Classification of Short-Duration Oceanic Signals," IEEE Journal of Oceanic Engineering, Vol. 17, No. 4, 1992.
....[Chakravarthy and Ghosh (1996) and the relevant results therein are summarized here. RBFN parameters can be calculated adaptively by minimizing the error in the network performance. This involves performing gradient descent on a quadratic error function E p , as given by the following equations: [Ghosh et al. 1992)] Deltaw ij = j 1 (t p i Gamma f i (xp ) R j (xp ) 1) 1 THIS WORK WAS SUPPORTED IN PART BY ARO CONTRACTS DAAH04 94 G 0417, DAAH04 9510494, AND THE TEXAS ADVANCED TECHNOLOGY PROGRAM GRANT ATP 442. Deltax j = j 2 R j (xp ) xp Gamma x j ) oe 2 j ( X i (t p i Gamma f i (xp ) w ij ....
Ghosh, J., Beck, S., Deuser, L., 1992, "A Neural Network Based Hybrid System for Detection, Characterization and Classification of Short-Duration Oceanic Signals," IEEE Jl. of Ocean Engineering, Vol. 17, pp. 351--363.
.... Gamma f i (x p ) 2 . Here t p i is the target function for input x p and f i is as defined in equation (1) The mean square error is the expected value of E p over all patterns. The parameters can be changed adaptively by performing gradient descent on E p as given by the following equations [24]: Deltaw ij = j 1 (t p i Gamma f i (x p ) R j (x p ) 2) Deltax j = j 2 R j (x p ) x p Gamma x j ) oe 2 j ( X i (t p i Gamma f i (x p ) w ij ) 3) Deltaoe j = j 3 R j (x p ) kx p Gamma x j k 2 oe 3 j ( X i (t p i Gamma f i (x p ) w ij ) 4) We will presently see ....
J. Ghosh, S. Beck, and L. Deuser, "A neural network based hybrid system for detection, characterization and classification of short-duration oceanic signals," IEEE Jl. of Ocean Engineering, vol. 17, pp. 351--363, October 1992.
No context found.
J. Ghosh, L. Deuser, and S. Beck, "A Neural Network based Hybrid System for Detection, Characterization and Classification of Short-duration Oceanic Signals," IEEE Journal of Ocean Engineering, vol. 17, no. 4, pp. 351--363, October 1992.
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