KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Data fusion
Process of integrating data from multiple sources to produce more consistent, accurate, and useful information than that provided by each individual source.
Pixel-level fusion
Multi-sensor fusion technique where raw data are directly combined at the most elementary level before any feature extraction or classification.
Feature-level fusion
Fusion method where features extracted from each sensor are combined before the final decision process, enabling dimensionality reduction and better interpretability.
Decision-level fusion
Fusion approach where each sensor performs its own processing before individual decisions are combined to produce a final consensus.
Heterogeneous sensors
Set of sensors with different technologies, resolutions, sampling rates, and measurement domains requiring harmonization for effective fusion.
Bayesian fusion
Probabilistic approach using Bayes' theorem to combine information from multiple sources by updating beliefs based on new observations.
Deep multimodal learning
Deep neural network architecture designed to simultaneously process and merge multiple different types of sensory data.
Sensor redundancy
Use of multiple sensors measuring the same physical quantity to improve robustness, detect failures, and increase measurement confidence.
Sensor complementarity
Strategy of combining sensors measuring different physical quantities to obtain a more comprehensive view of the monitored environment.
Real-time fusion
Fusion system capable of processing and combining sensor data with strict temporal constraints for critical applications.
Distributed architecture
System structure where data fusion is performed at different hierarchical levels, from the sensor to the central system.
State inference
Process of estimating the hidden state variables of a system from noisy observations coming from multiple sensors.
Multimodal cross-validation
Evaluation technique using validation data from different sensory modalities to test the robustness of the fusion model.
Predictive fusion
Fusion approach integrating predictive models to anticipate future system states and combine these predictions with current measurements.