Fully Integrated Analog Machine Learning Classifier Using Custom Activation Function for Low Resolution Image Classification

• Published in: IEEE transactions on circuits and systems. I, Regular papers Vol. 68; no. 3; pp. 1023 - 1033
• Main Authors: Tannirkulam Chandrasekaran, Sanjeev, Jayaraj, Akshay, Elkoori Ghantala Karnam, Vinay, Banerjee, Imon, Sanyal, Arindam
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.03.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; analog neural network; Classification; Classifiers; CMOS; Co-design; custom activation function; Datasets; Energy consumption; Energy costs; Energy resolution; Handwriting recognition; Image classification; Image resolution; low resolution image classification; Machine learning; Neural networks; Neurons; Pixels; Signal resolution; Training; Transistors
• ISSN: 1549-8328; 1558-0806
• DOI: 10.1109/TCSI.2020.3047331

This paper presents fully-integrated analog neural network classifier architecture for low resolution image classification that eliminates memory access. We design custom activation functions using single-stage common-source amplifiers, and apply a hardware-software co-design methodology to incorporate knowledge of the custom activation functions into the training phase to achieve high accuracy. Performing all computations entirely in the analog domain eliminates energy cost associated with memory access and data movement. We demonstrate our classifier on multinomial classification task of recognizing down-sampled handwritten digits from MNIST dataset. Fabricated in 65nm CMOS process, the measured energy consumption for down-sampled MNIST dataset is 173pJ/classification, which is <inline-formula> <tex-math notation="LaTeX">3\times </tex-math></inline-formula> better than state-of-the-art. The prototype IC achieves mean classification accuracy of 81.3% even after down-sampling the original MNIST images by 96% from <inline-formula> <tex-math notation="LaTeX">28\times 28 </tex-math></inline-formula> pixels to <inline-formula> <tex-math notation="LaTeX">5\times 5 </tex-math></inline-formula> pixels.