Open Access
Review
Table 3.
Function safety countermeasures at AI-empowered decision layer
Threats | Countermeasures | Remarks | Representative solutions |
---|---|---|---|
Optimization of aggregation algorithms | Federated learning | [167, 168] | |
Client anomaly detection | Federated learning | [169–171] | |
Low-quality unavailable | Decentralized approach | Federated learning | [172–174] |
computational outputs | Data pre-processing | Large language model | [175–178] |
Knowledge updating | Large language model | [179–181] | |
Alleviating hallucinations | Large language model | [182–184] | |
Explainable work | Large language model | [185, 186] | |
Asynchronous algorithms | Federated learning | [187, 188] | |
Efficient communications | Federated learning | [189–192] | |
Inefficient and overloaded | Resource management | Federated learning | [193–196] |
resource utilization | Efficient pre-training | Large language model | [110, 111] |
Parameter-efficient fine-tuning | Large language model | [197–199] | |
Efficient inference | Large language model | [200–202] | |
Environmentally friendly | Large language model | [111] |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.