A Low-Energy Machine-Learning Classifier Based on Clocked Comparators for Direct Inference on Analog Sensors
A Low-Energy Machine-Learning Classifier Based on Clocked Comparators for Direct Inference on Analog Sensors
复制标题
基于时钟比较器的低能耗机器学习分类器，用于模拟传感器的直接推理
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
N. Verma
中科院分区：
文献类型：
--
作者：
Zhuo Wang;N. Verma
本文介绍了一种系统，其中时钟比较器仅消耗<inline-formula><tex-math notation="LaTeX">$CV^{2}$</tex-math></inline-formula>能量，直接从模拟传感器信号中获得分类决策，从而取代仪表放大器、ADC和数字MAC（通常需要）。提出了一种用于训练分类器的机器学习算法，该算法能够克服模拟电路中的电路非理想性以及严重的能量/面积缩放。此外，噪声模型的系统提出和实验验证，提供了一种手段来预测和优化分类错误概率在给定的应用。噪声模型表明，与基于线性低噪声放大器的系统相比，基于比较器的系统实现了上级噪声效率。一个130纳米CMOS的原型通过将原始模拟像素作为输入来执行手写数字的图像识别。由于芯片上的引脚限制，28美元的图像<inline-formula><tex-math notation="LaTeX"> 对28 × 28=784 × $</tex-math></inline-formula>像素进行大小调整和下采样，以给出47个像素特征，对于理想的十路分类系统（MATLAB模拟），产生90%的准确度。基于原型比较器的系统实现了相同的性能，每十路分类的总能量为543 pJ，速度高达每秒130万张图像，相当于<inline-formula><tex-math notation="LaTeX">33美元 比</tex-math></inline-formula>ADC/数字MAC系统低10倍的能量。
This paper presents a system, where clocked comparators consuming only <inline-formula> <tex-math notation="LaTeX">$CV^{2}$ </tex-math></inline-formula> energy directly derive classification decisions from analog sensor signals, thereby replacing instrumentation amplifiers, ADCs, and digital MACs, as typically required. A machine-learning algorithm for training the classifier is presented, which enables circuit non-idealities as well as severe energy/area scaling in analog circuits to be overcome. Furthermore, a noise model of the system is presented and experimentally verified, providing a means to predict and optimize classification error probability in a given application. The noise model shows that superior noise efficiency is achieved by the comparator-based system compared with a system based on linear low-noise amplifiers. A prototype in 130-nm CMOS performs image recognition of handwritten numerical digits, by taking raw analog pixels as the inputs. Due to pin limitations on the chip, the images with <inline-formula> <tex-math notation="LaTeX">$28 imes 28=784$ </tex-math></inline-formula> pixels are resized and downsampled to give 47 pixel features, yielding an accuracy of 90% for an ideal ten-way classification system (MATLAB simulated). The prototype comparator-based system achieves equivalent performance with a total energy of 543 pJ per ten-way classification at a rate up to 1.3 M images per second, representing <inline-formula> <tex-math notation="LaTeX">$33 imes $ </tex-math></inline-formula> lower energy than an ADC/digital-MAC system.