AI 용어집
인공지능 완전 사전
Graph Neural Networks (GNN)
Deep learning architecture designed to process structured graph data, enabling the learning of node and edge representations through message propagation.
Traveling Salesman Problem (TSP)
NP-hard optimization problem seeking the minimum weight Hamiltonian cycle that visits each vertex exactly once in a complete weighted graph.
Minimum Vertex Cover
Minimal set of vertices such that every edge in the graph has at least one endpoint in this set, a fundamental combinatorial optimization problem.
Graph Coloring
Assignment of colors to vertices of a graph such that no two adjacent vertices share the same color, aiming to minimize the total number of colors used.
Reinforcement Learning on Graphs
Approach combining RL and graph structures where the agent makes decisions on nodes/edges to optimize a global objective on the graph topology.
Minimum Spanning Tree
Connected acyclic subgraph including all vertices with minimal sum of edge weights, efficiently solved by Kruskal's or Prim's algorithms.
Maximum Flow
Problem aiming to determine the maximum possible flow between a source and a sink in a directed graph with capacities on edges.
Community Detection
Identification of densely connected groups of nodes in a graph, using modularity metrics or unsupervised learning approaches.
Graph Centrality
Set of metrics evaluating the relative importance of nodes in a network, including degree, betweenness, closeness, and eigenvector centralities.
Graph Simulated Annealing
Optimization metaheuristic inspired by thermodynamics applied to graph problems, accepting degraded solutions with decreasing probability.
Genetic Algorithm for Graphs
Evolutionary approach where chromosomes represent graph solutions, using crossover and mutation to explore combinatorial solution spaces.
Multi-objective Optimization on Graphs
Simultaneous resolution of multiple conflicting objectives on graph structures, producing a Pareto front of non-dominated solutions.
Graph Isomorphism Problem
Determination of whether two graphs are structurally identical despite different vertex labeling, a key problem in complexity theory.
Maximum Cut Optimization
NP-hard problem seeking a vertex partition that maximizes the total weight of edges crossing the cut between the two sets.
Routing Algorithms in Graphs
Set of techniques determining optimal or near-optimal paths in networks, combining heuristics and learning for dynamic adaptation.
Spectral Clustering on Graphs
Partitioning method using eigenvectors of the graph Laplacian to project data into a space where clustering becomes trivial.