Training deep learning networks involves continuous weight updates across the various layers of the deep network while using a backpropagation (BP) algorithm. This results in expensive computation overheads during training. Consequently, most deep learning accelerators today employ pretrained weights and focus only on improving the design of the inference phase. The recent trend is to build a complete deep learning accelerator by incorporating the training module. Such efforts require an ultra-fast chip architecture for executing the BP algorithm. In this article, we propose a novel photonics-based backpropagation accelerator for high-performance deep learning training. We present the design for a convolutional neural network (CNN), BPLight-CNN , which incorporates the silicon photonics-based backpropagation accelerator. BPLight-CNN is a first-of-its-kind photonic and memristor-based CNN architecture for end-to-end training and prediction. We evaluate BPLight-CNN using a photonic CAD framework (IPKISS) on deep learning benchmark models, including LeNet and VGG-Net. The proposed design achieves (i) at least 34× speedup, 34× improvement in computational efficiency, and 38.5× energy savings during training; and (ii) 29× speedup, 31× improvement in computational efficiency, and 38.7× improvement in energy savings during inference compared with the state-of-the-art designs. All of these comparisons are done at a 16-bit resolution, and BPLight-CNN achieves these improvements at a cost of approximately 6% lower accuracy compared with the state-of-the-art.

中文翻译：

---

BPLight-CNN：用于深度学习的基于光子学的反向传播加速器

训练深度学习网络涉及在使用反向传播 (BP) 算法的同时跨深度网络的各个层进行连续权重更新。这会在训练期间导致昂贵的计算开销。因此，当今大多数深度学习加速器都使用预训练的权重，并且只专注于改进推理阶段的设计。最近的趋势是通过结合训练模块来构建一个完整的深度学习加速器。这样的努力需要一个超快的芯片架构来执行 BP 算法。在本文中，我们提出了一种新颖的基于光子学的反向传播加速器，用于高性能深度学习训练。我们提出了卷积神经网络 (CNN) 的设计，BPLight-CNN，它结合了基于硅光子学的反向传播加速器。BPLight-CNN是首创的基于光子和忆阻器的 CNN 架构，用于端到端训练和预测。我们评估BPLight-CNN在深度学习基准模型（包括 LeNet 和 VGG-Net）上使用光子 CAD 框架 (IPKISS)。所提出的设计实现了 (i) 至少 34 倍的加速，34 倍的计算效率提高，以及 38.5 倍的训练期间的能量节省；(ii) 与最先进的设计相比，在推理过程中加速了 29 倍，计算效率提高了 31 倍，节能提高了 38.7 倍。所有这些比较都是在 16 位分辨率下完成的，与最先进的技术相比，BPLight-CNN 以大约 6% 的精度降低成本实现了这些改进。