KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Horizontal Federated Learning
Configuration where clients share the same feature space but have different samples, typical in mobile applications with similar but distinct user data.
Vertical Federated Learning
Scenario where clients hold different features for the same set of samples, common in collaborations between companies with complementary user views.
Communication Round
Complete cycle of client local training followed by aggregation of updates to the central server, a fundamental unit of efficiency measurement in federated learning.
Non-IID Data
Distribution of data that is not identically and independently distributed among clients, a major challenge in federated learning that can cause convergence and model bias issues.
Federated Transfer Learning
Combination of transfer learning and federated learning to adapt pre-trained knowledge to specific tasks while preserving the privacy of local data.
Byzantine-Robust Federated Learning
Federated learning approaches resistant to malicious or faulty clients who may send incorrect updates to compromise the integrity of the global model.