Thuật ngữ AI
Từ điển đầy đủ về Trí tuệ nhân tạo
Particle Swarm
Set of particles collaboratively exploring the search space, where each particle adjusts its trajectory based on its own experience and that of its neighbors to converge toward an optimal solution.
Inertia Coefficient (w)
Scalar parameter controlling the influence of previous velocity on the current update, balancing global exploration and local exploitation within the PSO algorithm.
Cognitive Coefficient (c1)
Multiplicative weight applied to the attraction component toward the best personal position (pbest), regulating the importance of individual experience in the velocity update.
Social Coefficient (c2)
Multiplicative weight applied to the attraction component toward the global best position (gbest), controlling the influence of collective experience on particle behavior.
Random Factors (r1, r2)
Random numbers uniformly distributed in [0,1] introducing stochasticity in the cognitive and social components, ensuring trajectory diversification and avoiding premature convergence.
Star Topology
Communication structure where all particles are connected to a central particle (gbest), promoting fast convergence but risking trapping the algorithm in local optima.
Maximum Velocity (Vmax)
Absolute upper limit of the velocity vector magnitude, preventing particles from making excessive jumps in the search space and stabilizing algorithm behavior.
Velocity Update
Fundamental PSO equation combining inertia, cognitive attraction, and social attraction to calculate the new velocity of a particle in the next iteration.
Position Update
Additive equation where a particle's new position is obtained by adding its updated velocity to its current position, ensuring movement in the search space.