Neural computing and stochastic optimization
The potential of neural networks to find global optimum should be further explored. Their ability to do so using only local “gradient” information is surprising and can lead to very useful applications. Implementing an optimum receiver-decoder is a particularly interesting example.
Unable to display preview. Download preview PDF.
- [WON91b]E. Wong, “Implementing Boltzmann machines,” in Stochastic Analysis, E. Mayer-Wolf, E. Merzbach and A. Schwartz, eds., Academic Press, San Diego, 1991.Google Scholar