nsv2020.github.io - NSV 2020 - 13th International Workshop on Numerical Software Verification 2020

home > domains > nsv2020.github.io

Example domain paragraphs

Go to schedule»

The accuracy, energy, and latency of deep neural network (DNN) implementation is a strong function of the precision of its weights, activations and computation. Determining the minimum precision requirements of DNNs is made difficult by the inherent complexity of such networks. Hence, much of the work in literature is ad-hoc, e.g., pruning, binarization, and others. This talk address the problem of determining minimum precision requirements of DNNs with theoretical guarantees on its inference accuracy by se

Numerical validation enables one to improve the reliability of numerical computations that rely upon floating-point operations through obtaining trustful results. Discrete Stochastic Arithmetic (DSA) makes it possible to validate the accuracy of floating-point computations using random rounding. However, it may bring a large performance overhead compared with the standard floating-point operations. In this article, we show that with perturbed data it is possible to use standard floating-point arithmetic ins

Links to nsv2020.github.io (4)



© example3.com 2024