What is FSS?
FSS is a family of algorithms suited for optimization in high-dimensional search spaces. All fish perform local search and the school aggregates social information.
Why using FSS?
FSS is fast, it outperforms most other Swarm Intelligent algorithms and is computationally inexpensive. New investigations also show that FSS is easy to go-GPU.
Principles of FSS?
(i) Simple computations in all individuals (i.e. fish)
(ii) Various means of storing information (i.e. weights of fish and school barycenter)
(iii) Local computations (i.e. swimming is composed of distinct components)
(iv) Low communications between neighboring individuals (i.e. fish are to think local but also be socially aware)
(v) Minimum centralized control (mainly for self-controlling of the school radius)
(vi) Some distinct diversity mechanisms (this to avoid undesirable flocking behavior)
(vii) Scalability (in terms of complexity of the optimization/search tasks)
(viii) Autonomy (i.e. ability to self control functioning)
Rationale?
1. ‘Swimming’ actually is a means of:
- Performing a local search
- Storing information of success
- Indirectly conveying social information
2. Success of the search is given by:
- Fish weights (large is better)
- School radius (small is better – meaning more heavy fishes)
- School barycenter (closer to optima is better)
3. Non-monotonicity is achieved, e.g.:
- By random hesitation before swim
- By expansion/shrinking the school radius
- By variations on swimming components