Discussing the article: "The Chimp Optimization Algorithm: From ChOA to BChimp"

[MetaQuotes]

Check out the new article: The Chimp Optimization Algorithm: From ChOA to BChimp.

The Chimp Optimization Algorithm (ChOA) mimics primate group hunting with a division of roles, and its binary variant, BChimp, applies this mechanism to feature selection problems. We implement a continuous core in C_AO; along the way, we identify and correct an inherited coefficient defect — one that is masked by binarization but disrupts the search in the continuous domain. The abstract provides a ready-made implementation and practical insights into the quality and robustness of the search.

Chimpanzees live in a fission-fusion society, where the composition and size of hunting groups are constantly changing, and where hunting success depends not on uniformity but on the division of roles. The authors of the algorithm identify four types of individuals with different abilities and strategies: drivers follow the prey and pressure it without attacking; barriers block escape routes, trapping the prey; chasers quickly catch up with the prey, causing panic; and, finally, attackers anticipate the escape route and finish the hunt. In optimization terms, these four roles are assigned to the four best solutions in the population, and the remaining individuals update their positions based on all four leaders at once, rather than on a single leader. It is precisely this heterogeneity — the different laws governing changes in coefficients for different groups — that distinguishes ChOA from the homogeneous GWO family.

The second pillar of the algorithm is what the authors refer to as social motivation. In addition to rational hunting behavior, chimpanzees tend to exhibit impulsive, chaotic behavior — a drive to obtain meat at any cost for the sake of subsequent social exchange. In the mathematical model, this translates into a chaotic component: with a certain probability, an individual updates its position not according to the hunting formulas, but by a jump determined by a chaotic map. The purpose of this mechanism is to enable the algorithm to escape from local optima to which a smooth collective search inevitably converges.

Mathematically, ChOA can be described concisely. Each individual's position is recalculated as the average of four candidate positions. Each candidate is drawn toward its leader using a distance distorted by random and chaotic coefficients. The dynamic coefficient, which determines the balance between exploration and exploitation, decreases monotonically over the course of the iterations: initially, the steps are large and the population explores the space extensively; toward the end, it converges toward the leaders it has found. The chaotic replacement described above operates on top of this.

Author: Andrey Dik