AI 용어집
인공지능 완전 사전
Constraint-based Program Synthesis
Approach using logical constraints and SAT/SMT solvers to automatically generate programs satisfying specifications.
Neural Architecture Search by reinforcement
Using reinforcement learning agents to explore the space of neural network architectures and find optimal configurations.
Evolutionary Neural Architecture Search
Application of genetic and evolutionary algorithms to evolve and optimize neural network architectures.
Inductive program synthesis
Automatic program generation from input-output examples without explicit specifications.
Gradient-based Neural Architecture Search
Differentiable optimization of network architectures by relaxing the continuous search space to allow gradient descent.
Deductive program synthesis
Program generation through formal logical reasoning from mathematical specifications and preconditions/postconditions.
One-shot Neural Architecture Search
Efficient approaches training a single super-network enabling rapid evaluation of multiple architectures without retraining.
Probabilistic program synthesis
Program generation by modeling uncertainty and using probabilistic methods for search space exploration.
Architecture Search for quantum networks
Specialized optimization of quantum neural network architectures adapted to quantum computing constraints.
Synthèse de programmes par transformation
Génération de programmes par application systématique de règles de transformation sur des programmes existants ou des squelettes de code.
Neural Architecture Search multi-objectifs
Optimisation simultanée de plusieurs métriques comme la précision, la latence, la consommation énergétique et la taille du modèle.
Synthèse de programmes par apprentissage automatique
Utilisation de modèles d'apprentissage profond comme les transformers pour générer directement du code à partir de descriptions en langage naturel.