R. Battiti and G. Tecchiolli, "Learning with First, Second, and No Derivatives: a Case Study in High Energy Physics," Neur0computing, vol. 6, pp. 181 206, 1994. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
Hardware Learning in Analogue VLSI Neural Networks - Lehmann (1994) (4 citations) (Correct)
.... for an analogue VLSI implementation) Valle et al. 251] The disadvantage is that additional memory is needed (O(N 2 ) y Note, however, that on line learning never converges (for constant learning rate) the weights will stir about the optimal solution (White [264] Battiti and Tecchiolli [19]) Chapter 4.2 Implementation of on chip back propagation Page 62 4.2 Mapping the algorithm on VLSI The MLP recall mode equation (6 4 ) can be written y l = g(s l ) s l = w l z l , as noted in section 2.2 (see Lehmann and Bruun [143] Widrow and Lehr [265] Likewise, we can write ....
....in computing the activation function derivatives needed by gradient descent have inspired some authors to do without the derivative information; for instance by substituting a constanty for the derivative or by using completely different optimization techniques. See e.g. Battiti and Tecchiolli [19]; also Krogh et al. 125] We shall not elaborate on such solutions, as one of the objectives of this work is to approximate standard algorithms with known properties. y Assuming a monotonous activation function. Knowledge of the sign of the derivative should be sufficient for convergence (using ....
Roberto Battiti and Giampietro Tecchiolli, "Learning with first, second, and no derivatives: a case study in High Energy Physics," Neurocomputing, vol. 6, pp. 181-206, 1994.
Training Neural Nets with the Reactive Tabu Search - Battiff, Tecchiolli Self-citation (Tecchiolli) (Correct)
.... 64 23370.0 (77689.8) 20.0 32 89302.7 (158133.0) Available techniques to increase the afety requirements and the speed of convergence of BP, are, for exam ple, the use of adaptive learning rates for on line BP [30] the use of line searches and second order information for batch BP, see [7], 3] and the references contained. B, Event discrimination in High Encr9y Physics Experimental HEP facilities need state of the art discrimination systems for selecting and classifying the relevant events. In a typical facility, colliding particles produce streams of secondary particles ....
....events produced by the bottom quark as a benchmark for the T algorithm, both because of its applicative interest and because the large number of events available permits a statistically significant test of the relative performance of different algorithms. The same benchmark task has been used in [7] for comparing different training algorithms: i) the backpropagation algorithm [39] ii) a version of gradient descent with adaptive step, iii) the conjugate gradient technique, iv) the On Slcp Scan method with fast line searches [3] v) two versions of the stochastic search technique of [41] ....
[Article contains additional citation context not shown here]
R. Battiti and G. Tecchiolli, "Learning with First, Second, and No Derivatives: a Case Study in High Energy Physics," Neur0computing, vol. 6, pp. 181 206, 1994.
Training Neural Nets with the Reactive Tabu Search - Battiti, Tecchiolli (1995) (14 citations) Self-citation (Battiti Tecchiolli) (Correct)
.... (see also [22] for a discussion of cases where local minima are absent) Available techniques to increase the safety requirements and the speed of convergence of BP, are for example the use of variable step lengths according to heuristic criteria, and the use of secondorder information, see [8], 4] and the references contained. 4.2 Classification of Sonar Targets The three layer feed forward network used for classifying sonar returns from undersea metal cylinders or cylindrically shaped rocks is studied by Gorman and Sejnowski in [23] We consider this task as a benchmark for studying ....
....events produced by the bottom quark as a benchmark for the RTS algorithm, both because of its applicative interest and because the large number of events available permits a statistically significant test of the relative performance of di#erent algorithms. The same benchmark task has been used in [8] for comparing di#erent training algorithms: i) the backpropagation algorithm [41] ii) a version of gradient descent with adaptive step, iii) the conjugate gradient technique, iv) the One Step Secant method with fast line searches [4] v) two versions of the stochastic search technique of [43] ....
[Article contains additional citation context not shown here]
R. Battiti and G. Tecchiolli, "Learning with First, Second, and No Derivatives: a Case Study in High Energy Physics," Neurocomputing, in press, 1993.
Optimised Training Techniques For Feedforward Neural Networks - de Castro, Von Zuben (1998) (Correct)
No context found.
Battiti, R., "Learning with First, Second, and no Derivatives: A Case Study in High Energy Phisycs", Neurocomputing, NEUCOM 270, vol. 6, pp. 181-206, 1994, URL: ftp://ftp.cis.ohiostate. edu/pub/neuroprose/battiti.neuro-hep.ps.Z.
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