Abstract by Puneet Jain
Representation of Swarm Optimization using Bi-Partite Graphs
Solving an optimization problem has various heuristics involved, depending on the environment, the thing to optimize, and the agents involved in solving the problem. These details can be abstracted out by formulating the problem as a graph-based optimization, with the state of the graph and transition to the next state defining the system behavior. In particular, we use a bi-partite graph representation. This provides us with a generalization for scale-free network optimization and spatial optimization problems. The bi-partite graph transitions intrinsically represent the time/distance/hops and the randomness of the system, without knowing the actual structure of the problem. We aim to use these formalisms to solve optimization problems in spatial and non-spatial domains by incorporating the structure of the problem in the transition parameters, particularly the best-M-of-N problem on networks.