Recently, there are increasing efforts on advancing optical neural networks (ONNs), which bring significant advantages for machine learning (ML) in terms of power efficiency, parallelism, and computational speed. With the considerable benefits in computation speed and energy efficiency, there are significant interests in leveraging ONNs into medical sensing, security screening, drug detection, and autonomous driving. However, due to the challenge of implementing reconfigurability, deploying multi-task learning (MTL) algorithms on ONNs requires re-building and duplicating the physical diffractive systems, which significantly degrades the energy and cost efficiency in practical application scenarios. This work presents a novel ONNs architecture, namely, \textit{RubikONNs}, which utilizes the physical properties of optical systems to encode multiple feed-forward functions by physically rotating the hardware similarly to rotating a \textit{Rubik's Cube}. To optimize MTL performance on RubikONNs, two domain-specific physics-aware training algorithms \textit{RotAgg} and \textit{RotSeq} are proposed. Our experimental results demonstrate more than 4$\times$ improvements in energy and cost efficiency with marginal accuracy degradation compared to the state-of-the-art approaches.

翻译：近年来，光学神经网络（ONN）的研发日益受到关注，其在机器学习（ML）领域展现出显著的能效、并行性和计算速度优势。凭借计算速度与能量效率的巨大提升，ONN在医学传感、安防检测、毒品识别和自动驾驶领域具有重要应用前景。然而，由于可重构性的实现难题，当前在ONN上部署多任务学习（MTL）算法时需要重建和复制物理衍射系统，这严重影响了实际应用场景中的能耗与成本效率。本文提出一种新型ONN架构——鲁比克光学神经网络（RubikONNs），该架构利用光学系统的物理特性，通过类似旋转“鲁比克魔方”的硬件旋转操作来实现多个前馈功能的编码。为优化RubikONNs的多任务学习性能，我们提出了两种领域特定的物理感知训练算法：旋转聚合（RotAgg）和旋转序列（RotSeq）。实验结果表明，与现有最优方法相比，本方案在能量和成本效率上实现了超过4倍的提升，同时仅产生可忽略的精度损失。