Optical computer has actually become an effective method for high-speed and energy-efficient data processing. Diffractive optical networks, specifically, allow massive identical calculation via making use of easy organized stage masks and the proliferation of light. Nevertheless, one significant obstacle continues to be: systems learnt model-based simulations frequently stop working to execute efficiently in actual speculative setups, where imbalances, sound, and design errors are hard to record.
发布者：Dr.Durant，转转请注明出处：https://robotalks.cn/reinforcement-learning-accelerates-model-free-training-of-optical-ai-systems/
