{"id":"arxiv/2607.20153","url":"https://arxiv.org/abs/2607.20153","tags":["cs.LG","eess.SY","math.OC"],"score":0,"title":"Local Stability and Gaussian Smoothing of Quantized Neural Networks","domain":"cs.LG","source":"arxiv","authors":["Sergey Salishev","Anton Makarov","Oleg Granichin"],"reviews":0,"subject":"cs.LG","summary":"We study Gaussian averaging as a smooth surrogate for quantized neural models. Under bounded local oscillation, we derive a local dimension-dependent bound on |f-g|, linking Gaussian smoothing to the stability analysis of discontinuous networks. We compute closed-form Gaussian averages of the rectified linear unit (ReLU) and sign activation functions, and illustrate the mechanism on a high-dimensional binary perceptron, where layer-preactivation aggregation under an explicit quantization-noise surrogate yields the Gaussian envelope used in inference-side smoothing and training-side smooth…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-07-22","reactions":{}}