AI-ordlista
Den kompletta ordlistan över AI
Synchronous Parallel Annealing
Simulated annealing algorithm where parallel processes execute in perfect synchronization, exchanging information at predefined synchronization points to ensure coherence in exploration.
Asynchronous Parallel Annealing
Parallelized version of simulated annealing where processes work independently without strict synchronization, periodically communicating their best solutions to accelerate global convergence.
Exchange Parallel Annealing
Approach where multiple independent Markov chains evolve in parallel and periodically exchange their states or thermodynamic information to avoid local minima.
Divisional Parallel Annealing
Method that divides the search space into subdomains assigned to different processors, each applying independent simulated annealing on its partition of the problem.
Multi-trial Parallel Annealing
Algorithm that simultaneously evaluates multiple candidates at each temperature, allowing for broader neighborhood exploration and more robust selection of accepted transitions.
Master-Slave Parallel Annealing
Parallel architecture where a master processor coordinates several slaves performing neighborhood evaluations in parallel, centralizing acceptance/rejection decisions.
Cellular Parallel Annealing
Variant inspired by cellular automata where each cell executes local annealing in parallel, communicating only with its immediate neighbors to propagate information in the search space.
Island Parallel Annealing
Distributed model where isolated populations (islands) evolve independently with periodic migrations of individuals between islands to maintain genetic diversity in the search.
Hybrid Parallel Annealing
Combination of simulated annealing with other parallel metaheuristics like genetic algorithms or tabu search, leveraging the strengths of each approach for improved optimization.
Adaptive Parallel Annealing
Algorithm that dynamically adjusts parallelization and temperature parameters based on the quality of found solutions, optimizing the use of available computational resources.
Distributed Parallel Annealing
Implementation spanning multiple geographically distributed machines, using network communication protocols to synchronize states and share the best discovered solutions.
Cooperative Parallel Annealing
Approach where parallel agents actively share their knowledge about the optimization landscape, collectively building a better understanding of the search space.
Grid Parallel Annealing
Algorithms executed on grid computing infrastructures, leveraging heterogeneous distributed resources to solve very large-scale optimization problems.
Real-time Parallel Annealing
Variant adapted to strict time constraints, where parallelization ensures obtaining acceptable quality solutions within predetermined deadlines for critical applications.
GPU Parallel Annealing
Implementation optimized for graphics architectures (GPU), exploiting massive SIMT parallelism to evaluate thousands of neighbors simultaneously and massively accelerate convergence.
Quasi-Newtonian Parallel Annealing
Advanced hybridization combining parallel annealing with quasi-Newtonian gradient approximations, using local information to more effectively guide parallel exploration.