AI 용어집
인공지능 완전 사전
Online Learning
Paradigm where models learn sequentially from data streams arriving one by one
Catastrophic Forgetting
Phenomenon where neural networks drastically lose previously acquired knowledge when learning new tasks
Elastic Weight Consolidation
Regularization technique that identifies and protects important model weights during incremental learning
Replay Methods
Approaches based on selective repetition of old examples to preserve previous knowledge
Progressive Neural Networks
Architecture that dynamically adds new neural columns while maintaining lateral connections to previous ones
Memory-Augmented Neural Networks
Neural networks integrating an explicit external memory to store and retrieve past knowledge
Continual Meta-Learning
Meta-learning approaches that optimize the model's continuous adaptation capacity to new tasks
Continual Learning without Tasks
Continual learning paradigm where boundaries between tasks are not explicitly defined
Continuous Reinforcement
Application of continuous learning principles to reinforcement learning systems
Lifelong learning
Concept encompassing the continuous acquisition of knowledge throughout the operational lifetime of the system
Incremental Decision Trees
Decision tree algorithms capable of updating without complete reconstruction
Streaming Algorithms
Algorithms optimized for processing continuous data streams with memory constraints
Concept Drift Detection
Techniques to identify changes in data distribution over time
Knowledge Distillation
Process of transferring knowledge from an old model to a new one to preserve expertise
Parameter Regularization
Constraint methods on model parameters to limit forgetting during incremental learning
Episodic Memory Management
Selection and forgetting strategies in memory systems for continual learning
Continual Domain Adaptation
Progressive adaptation techniques to new domains without losing previous capabilities
Hierarchical Continual Reinforcement Learning
Combination of hierarchical structures and continual learning for complex tasks