FireNN: Neural Networks Reliability Evaluation on Hybrid Platforms

The growth of neural networks complexity has led to adopt of hardware-accelerators to cope with the computational power required by the new architectures. The possibility to adapt the network for different platforms enhanced the interests of safety-critical applications. The reliability evaluation of neural networks are still premature and requires platforms to measure the safety standards required by mission-critical applications. For this reason, the interest in studying the reliability of neural networks is growing. We propose a new approach for evaluating the resiliency of neural networks by using hybrid platforms. The approach relies on the reconfigurable hardware for emulating the target hardware platform and performing the fault injection process. The main advantage of the proposed approach is to involve the on-hardware execution of the neural network in the reliability analysis without any intrusiveness into the network algorithm and addressing specific fault models. The implementation of FireNN, the platform based on the proposed approach, is described in the paper. Experimental analyses are performed using fault injection on AlexNet. The analyses are carried out using the FireNN platform and the results are compared with the outcome of traditional software-level evaluations. Results are discussed considering the insight into the hardware level achieved using FireNN.