… faults affecting Quantized Neural Networks (QNNs) and how to effectively decrease their
effects in hardware accelerators. … accelerated error injection for quantized neural networks. The …

… We simulate a memristive accelerator similar to [9], wherein each layer of selected neural
network model is mapped to one or more 128×128 arrays and each memristive cell maintains …

… far from scalable in the envisioned scenario of a neural network trained once in the cloud and
… fault-tolerance of emerging neural network accelerators by leveraging the algorithmic error-…

… SNN fault tolerance by … neural hardware accelerator to efficiently support the proposed
technique. The experimental results show that, for a 900-neuron network with even a high fault …

… In this work, we explore applying redundancy only to the most critical elements within the
NN hardware accelerator, opening a new design space exploration to reduce the resource …

… a different problem compared to our work; our goal is to enhance the fault tolerance of the
given … ; only a few accelerators support this type of network architecture. Techniques have also …

… Abstract—Due to their growing popularity and computational cost, deep neural networks (…
This paper deals with the design of faulttolerant, systolic array based DNN accelerators for …

… This method, however, sacrifices the fault tolerance and … fault tolerance of SC, taking a
ternary neural network (TNN) as an example, we propose a parallel SC-based NN accelerator …

… based neural networks and propose a ReRAM-based accelerator with a dedicated pipeline
design for Attention-based Neural Networks. When considering cells with hard faults in …

… Fault Tolerance for CNN Accelerators This section provides a survey of fault tolerance for
CNN accelerators, and … of the neural networks and introduce hardware or time redundancy for …