Node co-activations as a means of error detection—Towards fault-tolerant neural networks
Context:: Machine learning has proved an efficient tool, but the systems need tools to mitigate risks during runtime. One approach is fault tolerance: detecting and handling errors before they cause harm. Objective:: This paper investigates whether rare co-activations – pairs of usually segregated n...
Main Authors: | Lalli Myllyaho, Jukka K. Nurminen, Tommi Mikkonen |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2022-09-01
|
Series: | Array |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2590005622000509 |
Similar Items
-
FaTEMa: A Framework for Multi-Layer Fault Tolerance in IoT Systems
by: Mário Melo, et al.
Published: (2021-10-01) -
Fault Tolerant Design, Error Detection and Correction /
by: Hatcher, Stephaine, author 639643, et al.
Published: (2012) -
Fault and Error Tolerance in Neural Networks: A Review
by: Cesar Torres-Huitzil, et al.
Published: (2017-01-01) -
Revisiting Symptom-Based Fault Tolerant Techniques against Soft Errors
by: Hwisoo So, et al.
Published: (2021-12-01) -
Quantum error correction and fault tolerant quantum computing /
by: 320532 Gaitan, Frank
Published: (2008)