Glossario IA
Il dizionario completo dell'Intelligenza Artificiale
Neural Attention Mechanism
Mechanism that allows the model to dynamically weight the importance of different examples from the support set during query classification.
Embedding Function
Neural function that transforms raw inputs into representation vectors in a space where semantic similarities can be measured.
K-nearest Neighbors
Classification algorithm that classifies a new instance according to the classes of its k nearest neighbors in the feature space, similar to the principle of Matching Networks.
Distance Metric
Mathematical function that quantifies the dissimilarity between two points in a space, used to compare embeddings in Matching Networks.
Cross-episode Training
Training technique where gradients are calculated on several consecutive episodes to improve the learning stability of the meta-model.
Test-time Adaptation
Ability of the model to quickly adapt to new tasks during inference using only examples from the support set without retraining.