AI 용어집
인공지능 완전 사전
Federated Aggregation Algorithms
Mathematical methods for combining local model weights into a single global model
Communication and Network Optimization
Techniques to minimize bandwidth and optimize exchanges between server and clients
Security and Adversarial Attacks
Protection against threats such as data poisoning and model inversion
Heterogeneity Management
Handling variations in computing power, data distribution, and connectivity
Non-IID Federated Learning
Management of non-identically distributed data across different clients
Model Personalization
Adapting global models to each user's specific preferences
Convergence and Optimization
Techniques to accelerate convergence and guarantee the stability of distributed training
Secure Aggregation
Cryptographic protocols to protect gradients during their transmission
Framework and Implementations
Software tools and libraries for deploying federated learning systems
Malicious Client Detection
Algorithms for identifying and isolating participants with abnormal behavior
Federated Reinforcement Learning
Application of reinforcement learning principles in a federated context
Split Learning
Variant where the model is split between client and server to reduce communication
Federated Evaluation Metrics
Specific indicators to measure the performance and fairness of federated models
Asynchronous Federated Learning
Approaches where clients update the global model without strict synchronization
Model Compression
Techniques to reduce model size before transmission in a federated environment