In a Nature Communications research study, scientists from China have actually established an error-aware probabilistic upgrade (EaPU) technique that lines up memristor equipment’s loud updates with semantic network training, lowering power usage by almost 6 orders of size versus GPUs while increasing precision on vision jobs. The research study confirms EaPU on 180 nm memristor varieties and large simulations.
发布者：Dr.Durant，转转请注明出处：https://robotalks.cn/new-memristor-training-method-slashes-ai-energy-use-by-six-orders-of-magnitude/
