gTds report presents a back-propagation algorithm t/tat varies the number of hidden units. 77ti,s" algorithm is expected to escape local minima and makes it no longer necessary to decide the number of hidden ttnits. We tested this algorithm on two examples. One was exclusive-OR learning and the othe
β¦ LIBER β¦
Geometrical interpretation of the back-propagation algorithm for the perceptron
β Scribed by Marco Budinich; Edoardo Milotti
- Publisher
- Elsevier Science
- Year
- 1992
- Tongue
- English
- Weight
- 528 KB
- Volume
- 185
- Category
- Article
- ISSN
- 0378-4371
No coin nor oath required. For personal study only.
π SIMILAR VOLUMES
Back-propagation algorithm which varies
β
Yoshio Hirose; Koichi Yamashita; Shimpei Hijiya
π
Article
π
1991
π
Elsevier Science
π
English
β 495 KB
The geometric interpretation of inversio
β
Wenping Wang; Barry Joe
π
Article
π
1995
π
Elsevier Science
π
English
β 1022 KB
Another interpretation of the EM algorit
β
Richard J. Hathaway
π
Article
π
1986
π
Elsevier Science
π
English
β 241 KB
Propagation equations for the set of nat
β
Philip J Greenberg
π
Article
π
1970
π
Elsevier Science
π
English
β 822 KB
A back-of-the-envelope algorithm for the
β
Frans Borremans; Dirk Tavernier
π
Article
π
1977
π
John Wiley and Sons
π
English
β 359 KB
## Abstract It is shown that one of the two conjugate solutions compatible with the AB part of an ABX system can easily be eliminated, at an early stage of the analysis, by considering intensity ratios in the X region.
Geometric algorithms for the conformatio
β
J. CortΓ©s; T. SimΓ©on; M. Remaud-SimΓ©on; V. Tran
π
Article
π
2004
π
John Wiley and Sons
π
English
β 384 KB
## Abstract The efficient filtering of unfeasible conformations would considerably benefit the exploration of the conformational space when searching for minimum energy structures or during molecular simulation. The most important conditions for filtering are the maintenance of molecular chain inte