From algorithms to devices: Enabling machine learning through ultra-low-power VLSI mixed-signal array processing

Siddharth Joshi, Chul Kim, Sohmyung Ha, Gert Cauwenberghs

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Machine learning and related statistical signal processing are expected to endow sensor networks with adaptive machine intelligence and greatly facilitate the Internet of Things. As such, architectures embedding adaptive and learning algorithms on-chip are oft-ignored by the design community and present a new set of design trade-offs. This review focuses on efficient implementation of mixed-signal matrix-vector multiplication as a central computational primitive enabling machine learning and statistical signal processing, with specific examples in spatial filtering for adaptive beamforming. We describe adaptive algorithms amenable for efficient implementation with such primitives in the presence of noise and analog variability. We also briefly highlight current trends in high-density integration in emerging memory device technologies and their use in high-dimensional adaptive computing.

Original languageEnglish (US)
Title of host publication38th Annual Custom Integrated Circuits Conference
Subtitle of host publicationA Showcase for Integrated Circuit Design in Silicon Hills, CICC 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Volume2017-April
ISBN (Electronic)9781509051915
DOIs
StatePublished - Jul 26 2017
Event38th Annual Custom Integrated Circuits Conference, CICC 2017 - Austin, United States
Duration: Apr 30 2017May 3 2017

Other

Other38th Annual Custom Integrated Circuits Conference, CICC 2017
CountryUnited States
CityAustin
Period4/30/175/3/17

Fingerprint

Array processing
Adaptive algorithms
Learning systems
Signal processing
Beamforming
Learning algorithms
Sensor networks
Data storage equipment
Internet of things

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Cite this

Joshi, S., Kim, C., Ha, S., & Cauwenberghs, G. (2017). From algorithms to devices: Enabling machine learning through ultra-low-power VLSI mixed-signal array processing. In 38th Annual Custom Integrated Circuits Conference: A Showcase for Integrated Circuit Design in Silicon Hills, CICC 2017 (Vol. 2017-April). [7993650] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CICC.2017.7993650

From algorithms to devices : Enabling machine learning through ultra-low-power VLSI mixed-signal array processing. / Joshi, Siddharth; Kim, Chul; Ha, Sohmyung; Cauwenberghs, Gert.

38th Annual Custom Integrated Circuits Conference: A Showcase for Integrated Circuit Design in Silicon Hills, CICC 2017. Vol. 2017-April Institute of Electrical and Electronics Engineers Inc., 2017. 7993650.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Joshi, S, Kim, C, Ha, S & Cauwenberghs, G 2017, From algorithms to devices: Enabling machine learning through ultra-low-power VLSI mixed-signal array processing. in 38th Annual Custom Integrated Circuits Conference: A Showcase for Integrated Circuit Design in Silicon Hills, CICC 2017. vol. 2017-April, 7993650, Institute of Electrical and Electronics Engineers Inc., 38th Annual Custom Integrated Circuits Conference, CICC 2017, Austin, United States, 4/30/17. https://doi.org/10.1109/CICC.2017.7993650
Joshi S, Kim C, Ha S, Cauwenberghs G. From algorithms to devices: Enabling machine learning through ultra-low-power VLSI mixed-signal array processing. In 38th Annual Custom Integrated Circuits Conference: A Showcase for Integrated Circuit Design in Silicon Hills, CICC 2017. Vol. 2017-April. Institute of Electrical and Electronics Engineers Inc. 2017. 7993650 https://doi.org/10.1109/CICC.2017.7993650
Joshi, Siddharth ; Kim, Chul ; Ha, Sohmyung ; Cauwenberghs, Gert. / From algorithms to devices : Enabling machine learning through ultra-low-power VLSI mixed-signal array processing. 38th Annual Custom Integrated Circuits Conference: A Showcase for Integrated Circuit Design in Silicon Hills, CICC 2017. Vol. 2017-April Institute of Electrical and Electronics Engineers Inc., 2017.
@inproceedings{20e90c81e16847d7b46bca674baef335,
title = "From algorithms to devices: Enabling machine learning through ultra-low-power VLSI mixed-signal array processing",
abstract = "Machine learning and related statistical signal processing are expected to endow sensor networks with adaptive machine intelligence and greatly facilitate the Internet of Things. As such, architectures embedding adaptive and learning algorithms on-chip are oft-ignored by the design community and present a new set of design trade-offs. This review focuses on efficient implementation of mixed-signal matrix-vector multiplication as a central computational primitive enabling machine learning and statistical signal processing, with specific examples in spatial filtering for adaptive beamforming. We describe adaptive algorithms amenable for efficient implementation with such primitives in the presence of noise and analog variability. We also briefly highlight current trends in high-density integration in emerging memory device technologies and their use in high-dimensional adaptive computing.",
author = "Siddharth Joshi and Chul Kim and Sohmyung Ha and Gert Cauwenberghs",
year = "2017",
month = "7",
day = "26",
doi = "10.1109/CICC.2017.7993650",
language = "English (US)",
volume = "2017-April",
booktitle = "38th Annual Custom Integrated Circuits Conference",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - From algorithms to devices

T2 - Enabling machine learning through ultra-low-power VLSI mixed-signal array processing

AU - Joshi, Siddharth

AU - Kim, Chul

AU - Ha, Sohmyung

AU - Cauwenberghs, Gert

PY - 2017/7/26

Y1 - 2017/7/26

N2 - Machine learning and related statistical signal processing are expected to endow sensor networks with adaptive machine intelligence and greatly facilitate the Internet of Things. As such, architectures embedding adaptive and learning algorithms on-chip are oft-ignored by the design community and present a new set of design trade-offs. This review focuses on efficient implementation of mixed-signal matrix-vector multiplication as a central computational primitive enabling machine learning and statistical signal processing, with specific examples in spatial filtering for adaptive beamforming. We describe adaptive algorithms amenable for efficient implementation with such primitives in the presence of noise and analog variability. We also briefly highlight current trends in high-density integration in emerging memory device technologies and their use in high-dimensional adaptive computing.

AB - Machine learning and related statistical signal processing are expected to endow sensor networks with adaptive machine intelligence and greatly facilitate the Internet of Things. As such, architectures embedding adaptive and learning algorithms on-chip are oft-ignored by the design community and present a new set of design trade-offs. This review focuses on efficient implementation of mixed-signal matrix-vector multiplication as a central computational primitive enabling machine learning and statistical signal processing, with specific examples in spatial filtering for adaptive beamforming. We describe adaptive algorithms amenable for efficient implementation with such primitives in the presence of noise and analog variability. We also briefly highlight current trends in high-density integration in emerging memory device technologies and their use in high-dimensional adaptive computing.

UR - http://www.scopus.com/inward/record.url?scp=85030479238&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85030479238&partnerID=8YFLogxK

U2 - 10.1109/CICC.2017.7993650

DO - 10.1109/CICC.2017.7993650

M3 - Conference contribution

AN - SCOPUS:85030479238

VL - 2017-April

BT - 38th Annual Custom Integrated Circuits Conference

PB - Institute of Electrical and Electronics Engineers Inc.

ER -