LATIDIA · Ciberseguridad
La binarización como superficie de ataque en redes neuronales ligeras
arXiv:2611.17646v1 Tipo de anuncio: nuevo Resumen: Las redes neuronales sin peso (WNN) son clasificadores ligeros que binarizan sus entradas y aprenden almacenando patrones binarios en memorias de consulta, lo que las hace atractivas.
WhatsApp ↗Telegram ↗
La noticia
arXiv:2611.17646v1 Announce Type: new Abstract: Weightless Neural Networks (WNNs) are lightweight classifiers that binarize their inputs and learn by storing binary patterns in lookup memories, which makes them attractive for resource-constrained tasks such as intrusion detection. Because a small perturbation that crosses no binarization threshold leaves the binary input unchanged, WNNs may appear naturally robust to adversarial examples; a perturbation that does cross a threshold, however, can still change the prediction. We study this tension in WiSARD, the canonical RAM-based WNN, and in two variants that replace its memories with Bloom filters, Bloom WiSARD and BTHOWeN. A continuous perturbation reaches these classifiers only by crossing an encoding threshold, changing at