{"id":"arxiv/2607.04531","url":"https://arxiv.org/abs/2607.04531","tags":["cs.LG","cs.AI","eess.SP","math.NA"],"score":0,"title":"Lyapunov-Guided Training for Hardware-Safe Neural Networks Under Fixed-Point Arithmetic","domain":"cs.LG","source":"arxiv","authors":["Anis Hamadouche","Amir Hussain"],"reviews":0,"subject":"cs.LG","summary":"Low-precision neural networks are attractive for resource-constrained hardware, but fixed-point arithmetic introduces failure modes that are often hidden by idealised quantisation models. In particular, two's-complement overflow wrapping can corrupt hidden activations by changing both their magnitude and sign, leading to unstable numerical error propagation and severe accuracy degradation. This paper proposes a Lyapunov-stabilised quantisation framework for low-precision neural networks operating under hardware-style wrapping arithmetic. The hidden-state energy is monitored through a…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-07-05","reactions":{}}