Approximate Computing for ML: State-of-the-art, Challenges and Visions

Georgios Zervakis; Hassaan Saadat; Hussam Mrouch; Andreas Gerstlauer; Sri Parameswaran; Jorg Henkel

doi:10.1145/3394885.3431632

Approximate Computing for ML: State-of-the-art, Challenges and Visions

Georgios Zervakis, Hassaan Saadat, Hussam Mrouch, Andreas Gerstlauer, Sri Parameswaran, Jorg Henkel

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

31 Scopus citations

Abstract

In this paper, we present our state-of-the-art approximate techniques that cover the main pillars of approximate computing research. Our analysis considers both static and reconfigurable approximation techniques as well as operation-specific approximate components (e.g., multipliers) and generalized approximate highlevel synthesis approaches. As our application target, we discuss the improvements that such techniques bring on machine learning and neural networks. In addition to the conventionally analyzed performance and energy gains, we also evaluate the improvements that approximate computing brings in the operating temperature.

Original language	English
Title of host publication	Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	189-196
Number of pages	8
ISBN (Electronic)	9781450379991
DOIs	https://doi.org/10.1145/3394885.3431632
State	Published - 18 Jan 2021
Externally published	Yes
Event	26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021 - Virtual, Online, Japan Duration: 18 Jan 2021 → 21 Jan 2021

Publication series

Name	Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC

Conference

Conference	26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021
Country/Territory	Japan
City	Virtual, Online
Period	18/01/21 → 21/01/21

Keywords

Accelerator
Approximate Computing
Architecture
High-Level Synthesis
Inference
Logic
Low-power
Multiplier
Neural Network
Renconfigurable Accuracy
Temperature

Access to Document

10.1145/3394885.3431632

Cite this

Zervakis, G., Saadat, H., Mrouch, H., Gerstlauer, A., Parameswaran, S., & Henkel, J. (2021). Approximate Computing for ML: State-of-the-art, Challenges and Visions. In Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021 (pp. 189-196). Article 3431632 (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1145/3394885.3431632

Zervakis, Georgios ; Saadat, Hassaan ; Mrouch, Hussam et al. / Approximate Computing for ML : State-of-the-art, Challenges and Visions. Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 189-196 (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC).

@inproceedings{eb043c3485ee4b2aa689b80126ed0165,

title = "Approximate Computing for ML: State-of-the-art, Challenges and Visions",

abstract = "In this paper, we present our state-of-the-art approximate techniques that cover the main pillars of approximate computing research. Our analysis considers both static and reconfigurable approximation techniques as well as operation-specific approximate components (e.g., multipliers) and generalized approximate highlevel synthesis approaches. As our application target, we discuss the improvements that such techniques bring on machine learning and neural networks. In addition to the conventionally analyzed performance and energy gains, we also evaluate the improvements that approximate computing brings in the operating temperature.",

keywords = "Accelerator, Approximate Computing, Architecture, High-Level Synthesis, Inference, Logic, Low-power, Multiplier, Neural Network, Renconfigurable Accuracy, Temperature",

author = "Georgios Zervakis and Hassaan Saadat and Hussam Mrouch and Andreas Gerstlauer and Sri Parameswaran and Jorg Henkel",

note = "Publisher Copyright: {\textcopyright} 2021 Association for Computing Machinery.; 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021 ; Conference date: 18-01-2021 Through 21-01-2021",

year = "2021",

month = jan,

day = "18",

doi = "10.1145/3394885.3431632",

language = "English",

series = "Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "189--196",

booktitle = "Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021",

}

Zervakis, G, Saadat, H, Mrouch, H, Gerstlauer, A, Parameswaran, S & Henkel, J 2021, Approximate Computing for ML: State-of-the-art, Challenges and Visions. in Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021., 3431632, Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC, Institute of Electrical and Electronics Engineers Inc., pp. 189-196, 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021, Virtual, Online, Japan, 18/01/21. https://doi.org/10.1145/3394885.3431632

Approximate Computing for ML: State-of-the-art, Challenges and Visions. / Zervakis, Georgios; Saadat, Hassaan; Mrouch, Hussam et al.
Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021. Institute of Electrical and Electronics Engineers Inc., 2021. p. 189-196 3431632 (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Approximate Computing for ML

T2 - 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021

AU - Zervakis, Georgios

AU - Saadat, Hassaan

AU - Mrouch, Hussam

AU - Gerstlauer, Andreas

AU - Parameswaran, Sri

AU - Henkel, Jorg

PY - 2021/1/18

Y1 - 2021/1/18

N2 - In this paper, we present our state-of-the-art approximate techniques that cover the main pillars of approximate computing research. Our analysis considers both static and reconfigurable approximation techniques as well as operation-specific approximate components (e.g., multipliers) and generalized approximate highlevel synthesis approaches. As our application target, we discuss the improvements that such techniques bring on machine learning and neural networks. In addition to the conventionally analyzed performance and energy gains, we also evaluate the improvements that approximate computing brings in the operating temperature.

AB - In this paper, we present our state-of-the-art approximate techniques that cover the main pillars of approximate computing research. Our analysis considers both static and reconfigurable approximation techniques as well as operation-specific approximate components (e.g., multipliers) and generalized approximate highlevel synthesis approaches. As our application target, we discuss the improvements that such techniques bring on machine learning and neural networks. In addition to the conventionally analyzed performance and energy gains, we also evaluate the improvements that approximate computing brings in the operating temperature.

KW - Accelerator

KW - Approximate Computing

KW - Architecture

KW - High-Level Synthesis

KW - Inference

KW - Logic

KW - Low-power

KW - Multiplier

KW - Neural Network

KW - Renconfigurable Accuracy

KW - Temperature

UR - http://www.scopus.com/inward/record.url?scp=85100578547&partnerID=8YFLogxK

U2 - 10.1145/3394885.3431632

DO - 10.1145/3394885.3431632

M3 - Conference contribution

AN - SCOPUS:85100578547

T3 - Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC

SP - 189

EP - 196

BT - Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 18 January 2021 through 21 January 2021

ER -

Zervakis G, Saadat H, Mrouch H, Gerstlauer A, Parameswaran S, Henkel J. Approximate Computing for ML: State-of-the-art, Challenges and Visions. In Proceedings of the 26th Asia and South Pacific Design Automation Conference, ASP-DAC 2021. Institute of Electrical and Electronics Engineers Inc. 2021. p. 189-196. 3431632. (Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC). doi: 10.1145/3394885.3431632

Approximate Computing for ML: State-of-the-art, Challenges and Visions

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this