MILR: Mathematically Induced Layer Recovery for Plaintext Space Error Correction of CNNs
This addresses the need for robust and resilient CNNs in mission-critical systems, offering a novel software solution for error correction.
The paper tackles the problem of unreliable CNN inference due to faults or attacks by proposing MILR, a software-based error detection and correction system that enables self-healing from single and multi-bit errors, achieving recovery of erroneous weights and layers without hardware or network modifications.
The increased use of Convolutional Neural Networks (CNN) in mission critical systems has increased the need for robust and resilient networks in the face of both naturally occurring faults as well as security attacks. The lack of robustness and resiliency can lead to unreliable inference results. Current methods that address CNN robustness require hardware modification, network modification, or network duplication. This paper proposes MILR a software based CNN error detection and error correction system that enables self-healing of the network from single and multi bit errors. The self-healing capabilities are based on mathematical relationships between the inputs,outputs, and parameters(weights) of a layers, exploiting these relationships allow the recovery of erroneous parameters (weights) throughout a layer and the network. MILR is suitable for plaintext-space error correction (PSEC) given its ability to correct whole-weight and even whole-layer errors in CNNs.