Embedded devices are generally small, battery-powered computers with limited hardware resources. It is difficult to run deep neural networks (DNNs) on these devices, because DNNs perform millions of operations and consume significant amounts of energy. Prior research has shown that a considerable number of a DNN’s memory accesses and computation are redundant when performing tasks like image classification. To reduce this redundancy and thereby reduce the energy consumption of DNNs, we introduce the Modular Neural Network Tree architecture. Instead of using one large DNN for the classifier, this architecture uses multiple smaller DNNs (called modules ) to progressively classify images into groups of categories based on a novel visual similarity metric. Once a group of categories is selected by a module, another module then continues to distinguish among the similar categories within the selected group. This process is repeated over multiple modules until we are left with a single category. The computation needed to distinguish dissimilar groups is avoided, thus reducing redundant operations, memory accesses, and energy. Experimental results using several image datasets reveal the effectiveness of our proposed solution to reduce memory requirements by 50% to 99%, inference time by 55% to 95%, energy consumption by 52% to 94%, and the number of operations by 15% to 99% when compared with existing DNN architectures, running on two different embedded systems: Raspberry Pi 3 and Raspberry Pi Zero.

中文翻译：

---

用于嵌入式设备上的低功耗图像分类的模块化神经网络

嵌入式设备通常是小型、电池供电的计算机，硬件资源有限。在这些设备上运行深度神经网络 (DNN) 很困难，因为 DNN 执行数百万次操作并消耗大量能量。先前的研究表明，在执行图像分类等任务时，相当多的 DNN 内存访问和计算是多余的。为了减少这种冗余，从而降低 DNN 的能耗，我们引入了模块化神经网络树架构。该架构不是使用一个大的 DNN 作为分类器，而是使用多个较小的 DNN（称为模块) 根据一种新的视觉相似度度量逐步将图像分类为类别组。一旦一个模块选择了一组类别，另一个模块就会继续在所选组内的相似类别之间进行区分。这个过程在多个模块上重复，直到我们只剩下一个类别。避免了区分不同组所需的计算，从而减少了冗余操作、内存访问和能量。使用多个图像数据集的实验结果表明，我们提出的解决方案的有效性可以将内存需求减少 50% 到 99%，推理时间减少 55% 到 95%，能耗减少 52% 到 94%，操作次数减少 15%与现有的 DNN 架构相比，高达 99%，运行在两个不同的嵌入式系统上：Raspberry Pi 3 和 Raspberry Pi Zero。