A representation method for PWL functions oriented to parallel processing
✍ Scribed by J.M. Tarela; E. Alonso; M.V. Martínez
- Book ID
- 103928537
- Publisher
- Elsevier Science
- Year
- 1990
- Tongue
- English
- Weight
- 645 KB
- Volume
- 13
- Category
- Article
- ISSN
- 0895-7177
No coin nor oath required. For personal study only.
✦ Synopsis
PWL functions are represented by an algebra that facilitates their treatment in many practical problems. One of the characteristics of the resulting analytical model is that the representation of the data flow proves particularly suitable for parallel processing. The present work shows that the number of computation cells is lower than the number of hyperplanes of the function. Also the analytical expression does not require any information about the regions into which the function domain is divided. These two facts confer the offered method great usefulness within the context of neural computation. The authors thank Prof. J.I. Maeztu for his helpful comments on Lattice Theory. This paper was supported by the University of PaIs Vasco (projects UPVSlO.OS-S/86 and UPV139.310-0041/88).
📜 SIMILAR VOLUMES