Mafunzo Dhidi ya Mashambulizi
Mafunzo Dhidi ya Mashambulizi ni utaratibu thabiti wa uboreshaji kwa mitandao ya neva ya kina ambapo modeli hufunzwa sio tu kwa data safi bali pia kwa pembejeo zilizoathiriwa na hali mbaya zaidi zilizoundwa wakati wa mafunzo. Kwa kuunda mfumo rasmi na Madry et al. (2018) kama tatizo la kiwango cha juu-chini cha min-max, njia hutumia Projected Gradient Descent (PGD) kutoa mifano dhihirishi yenye nguvu ndani ya seti ya kikomo ya usumbufu wa Lp kabla ya kila sasisho la mteremko, ikilazimisha mtandao kujifunza mipaka ya uamuzi ambayo ni thabiti chini ya usumbufu kama huo.
Soma mbinu kamili
Ingia kwa akaunti ya bure ili kusoma sehemu hii.
Method map
The neighbourhood of related methods — select a node to explore.
Vyanzo
- Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2018). Towards deep learning models resistant to adversarial attacks. International Conference on Learning Representations (ICLR). link ↗
Jinsi ya kunukuu ukurasa huu
ScholarGate. (2026, June 2). Adversarial Training (Robust Optimization for DL). ScholarGate. https://scholargate.app/sw/deep-learning/adversarial-training
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Uongezaji DataUjifunzaji wa Kina↔ compare
- Mtandao wa Kushawishi unaozalisha (Generative Adversarial Network - GAN)Ujifunzaji wa Kina↔ compare
- Utambuzi wa Data Nje ya UsambazajiUjifunzaji wa Mashine↔ compare
Imerejelewa na
Umeona tatizo kwenye ukurasa huu? Ripoti au pendekeza marekebisho →