next up previous contents
Next: Other Variants and Up: Mean Field Annealing Previous: MFA variables: from

Conclusion

In conclusion, we would like to briefly state a few interesting points about neural algorithms. Neural networks differ from most of the other methods talked about in this paper, in that whereas those other methods are restricted to searching within a finite set of feasible solutions, the neural techniques explore more or less a continuous search space with an objective for picking out a solution. Another feature of the neural networks is that most of them are inherently parallel in operation. They are made up of a large number of simple elements that can process their inputs in parallel. Accordingly, they lend themselves naturally to implementations on parallel computers.



HPF demo web account
Mon Nov 18 19:45:42 EST 1996