KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Episodic Memory
Neural storage structure that preserves specific experiences contextualized in time, enabling AI systems to recall past events to guide future decisions.
Selective Replay
Targeted selection strategy for past experiences based on their informational importance, diversity, or rarity to optimize memory space and learning efficiency.
Forgetting Mechanism
Controlled selective forgetting system that progressively eliminates less relevant information from episodic memory to maintain optimal storage capacity.
Episodic Memory Network
Specialized neural architecture for encoding, storing, and retrieving specific experience episodes, integrating attention mechanisms for efficient selection.
Meta-Learning with Episodic Memory
Learning approach where episodic memory serves as meta-data to accelerate adaptation to new tasks by exploiting similarities with past experiences.
Episodic Memory Compression
Dimensionality reduction technique applied to stored experiences, preserving essential information while optimizing available memory space.
Memory Replay Prioritization
Weighted sampling algorithm that favors experiences with high prediction errors or informational surprise, thereby accelerating learning convergence.
Continual Learning with Episodic Memory
Learning paradigm where episodic memory enables sequential acquisition of skills without performance degradation on previously learned tasks.
Episodic Knowledge Distillation
Process of transferring knowledge from episodic memory to model parameters, preserving important patterns while compacting the representation.
Memory Sampling Strategies
Set of probabilistic techniques determining the selection of episodes from episodic memory for training, balancing exploration and exploitation of experiences.
Episodic Replay Scheduling
Temporal policy regulating the frequency and sequence of episodic reactivations to optimize learning consolidation and minimize interference.
Catastrophic Forgetting Prevention
Set of mechanisms using episodic memory to counter catastrophic forgetting during sequential learning of multiple tasks.
Episodic Memory Capacity Management
Dynamic system for allocating and reallocating episodic memory space based on the importance and recency of stored experiences.
Temporal Episodic Memory
Memory structure that explicitly encodes temporal relationships between episodes, enabling the system to understand sequences and chronological dependencies.
Hierarchical Episodic Memory
Multi-level architecture organizing experiences into hierarchical abstractions, from specific detail to general schema, for efficient retrieval and improved generalization.