Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond

Siva Satyendra Sahoo; Bastien Deveautour; Marcello Traiola; Chongyan Gu; Yun Wu; Aditya Japa; Salim Ullah; Akash Kumar

doi:10.1145/3742872.3757056

Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond

Siva Satyendra Sahoo, Bastien Deveautour, Marcello Traiola, Chongyan Gu, Yun Wu, Aditya Japa, Salim Ullah, Akash Kumar

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

Abstract

The rapid growth of AI workloads is driving interest in Approximate Computing (AxC) as a means to enable low-cost, energy-efficient inference in resource-constrained systems. By introducing controlled inaccuracies, AxC can deliver substantial gains in power, performance, and area (PPA) while leveraging the inherent error tolerance of many AI models. Achieving this potential requires adapting existing frameworks to support the design and optimization of neural networks with approximate operators. Modern AxC research extends beyond accuracy-PPA trade-offs to address reliability and security, reducing redundancy overheads and exploring the distinctive side-channel implications of approximation. Application-aware approaches, such as those for spiking neural networks, show that tailoring approximation to workload-specific error behavior can surpass generic strategies. This article examines AI-guided design methods and the interplay between efficiency, reliability, and security, highlighting how these interconnected facets can advance embedded and high-performance computing.

Original language	English
Title of host publication	CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems
Publisher	Association for Computing Machinery
Pages	11-20
Number of pages	10
ISBN (Print)	9798400719912
DOIs	https://doi.org/10.1145/3742872.3757056
Publication status	Published online - 28 Sept 2025
Event	CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems - Taipei, Taiwan Duration: 28 Sept 2025 → 3 Oct 2025

Publication series

Name	Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems
Publisher	Association for Computing Machinery

Conference

Conference	CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems
City	Taipei, Taiwan
Period	28/09/25 → 3/10/25

Bibliographical note

Funding

We acknowledge financial support from the following: Deutsche Forschungsgemeinschaft (DFG) under the X-ReAp project (Project number 380524764); The Conseil régional des Pays de la Loire, Nantes Université and the Institut d’Electronique et des Technologies du numéRique under the PULSAR project; Agence Nationale de la Recherche (ANR) under the RE-TRUSTING project, ANR-21-CE24-0015; EPSRC (UK) under the Grant EP/X009602/1.

Keywords

Approximate Computing
Reliability
Security
Energy
Power/Energy
Edge Computing
Edge AI
Fault Tolerance
Side-channel Attacks
fault tolerance
approximate computing
edge computing
reliability
security
side-channel attacks
power/energy
edge AI
energy

Access to Document

10.1145/3742872.3757056

Cite this

Sahoo, S. S., Deveautour, B., Traiola, M., Gu, C., Wu, Y., Japa, A., Ullah, S., & Kumar, A. (2025). Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond. In CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems (pp. 11-20). (Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems). Association for Computing Machinery. Advance online publication. https://doi.org/10.1145/3742872.3757056

Sahoo, Siva Satyendra ; Deveautour, Bastien ; Traiola, Marcello et al. / Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond. CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems. Association for Computing Machinery, 2025. pp. 11-20 (Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems).

@inproceedings{ba9871078a0840ecb212e8ae9ae329e8,

title = "Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond",

abstract = "The rapid growth of AI workloads is driving interest in Approximate Computing (AxC) as a means to enable low-cost, energy-efficient inference in resource-constrained systems. By introducing controlled inaccuracies, AxC can deliver substantial gains in power, performance, and area (PPA) while leveraging the inherent error tolerance of many AI models. Achieving this potential requires adapting existing frameworks to support the design and optimization of neural networks with approximate operators. Modern AxC research extends beyond accuracy-PPA trade-offs to address reliability and security, reducing redundancy overheads and exploring the distinctive side-channel implications of approximation. Application-aware approaches, such as those for spiking neural networks, show that tailoring approximation to workload-specific error behavior can surpass generic strategies. This article examines AI-guided design methods and the interplay between efficiency, reliability, and security, highlighting how these interconnected facets can advance embedded and high-performance computing.",

keywords = "Approximate Computing, Reliability, Security, Energy, Power/Energy, Edge Computing, Edge AI, Fault Tolerance, Side-channel Attacks, fault tolerance, approximate computing, edge computing, reliability, security, side-channel attacks, power/energy, edge AI, energy",

author = "Sahoo, \{Siva Satyendra\} and Bastien Deveautour and Marcello Traiola and Chongyan Gu and Yun Wu and Aditya Japa and Salim Ullah and Akash Kumar",

note = "Copyright {\textcopyright} 2025 Copyright is held by the owner/author(s). Publication rights licensed to ACM.; CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems ; Conference date: 28-09-2025 Through 03-10-2025",

year = "2025",

month = sep,

day = "28",

doi = "10.1145/3742872.3757056",

language = "English",

isbn = "9798400719912",

series = "Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems",

publisher = "Association for Computing Machinery",

pages = "11--20",

booktitle = "CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems",

address = "United States",

}

Sahoo, SS, Deveautour, B, Traiola, M, Gu, C, Wu, Y, Japa, A, Ullah, S & Kumar, A 2025, Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond. in CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems. Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems, Association for Computing Machinery, pp. 11-20, CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems, Taipei, Taiwan, 28/09/25. https://doi.org/10.1145/3742872.3757056

Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond. / Sahoo, Siva Satyendra; Deveautour, Bastien; Traiola, Marcello et al.
CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems. Association for Computing Machinery, 2025. p. 11-20 (Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems).

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

TY - GEN

T1 - Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond

AU - Sahoo, Siva Satyendra

AU - Deveautour, Bastien

AU - Traiola, Marcello

AU - Gu, Chongyan

AU - Wu, Yun

AU - Japa, Aditya

AU - Ullah, Salim

AU - Kumar, Akash

PY - 2025/9/28

Y1 - 2025/9/28

N2 - The rapid growth of AI workloads is driving interest in Approximate Computing (AxC) as a means to enable low-cost, energy-efficient inference in resource-constrained systems. By introducing controlled inaccuracies, AxC can deliver substantial gains in power, performance, and area (PPA) while leveraging the inherent error tolerance of many AI models. Achieving this potential requires adapting existing frameworks to support the design and optimization of neural networks with approximate operators. Modern AxC research extends beyond accuracy-PPA trade-offs to address reliability and security, reducing redundancy overheads and exploring the distinctive side-channel implications of approximation. Application-aware approaches, such as those for spiking neural networks, show that tailoring approximation to workload-specific error behavior can surpass generic strategies. This article examines AI-guided design methods and the interplay between efficiency, reliability, and security, highlighting how these interconnected facets can advance embedded and high-performance computing.

AB - The rapid growth of AI workloads is driving interest in Approximate Computing (AxC) as a means to enable low-cost, energy-efficient inference in resource-constrained systems. By introducing controlled inaccuracies, AxC can deliver substantial gains in power, performance, and area (PPA) while leveraging the inherent error tolerance of many AI models. Achieving this potential requires adapting existing frameworks to support the design and optimization of neural networks with approximate operators. Modern AxC research extends beyond accuracy-PPA trade-offs to address reliability and security, reducing redundancy overheads and exploring the distinctive side-channel implications of approximation. Application-aware approaches, such as those for spiking neural networks, show that tailoring approximation to workload-specific error behavior can surpass generic strategies. This article examines AI-guided design methods and the interplay between efficiency, reliability, and security, highlighting how these interconnected facets can advance embedded and high-performance computing.

KW - Approximate Computing

KW - Reliability

KW - Security

KW - Energy

KW - Power/Energy

KW - Edge Computing

KW - Edge AI

KW - Fault Tolerance

KW - Side-channel Attacks

KW - fault tolerance

KW - approximate computing

KW - edge computing

KW - reliability

KW - security

KW - side-channel attacks

KW - power/energy

KW - edge AI

KW - energy

UR - https://pure.ulster.ac.uk/en/publications/ba987107-8a08-40ec-b212-e8ae9ae329e8

UR - https://www.scopus.com/pages/publications/105025377544

U2 - 10.1145/3742872.3757056

DO - 10.1145/3742872.3757056

M3 - Conference contribution

SN - 9798400719912

T3 - Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems

SP - 11

EP - 20

BT - CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems

PB - Association for Computing Machinery

T2 - CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems

Y2 - 28 September 2025 through 3 October 2025

ER -

Sahoo SS, Deveautour B, Traiola M, Gu C, Wu Y, Japa A et al. Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond. In CASES '25: International Conference on Compilers, Architecture, and Synthesis for Embedded Systems. Association for Computing Machinery. 2025. p. 11-20. (Proceedings of the International Conference on Compilers, Architecture, and Synthesis for Embedded Systems). Epub 2025 Sept 28. doi: 10.1145/3742872.3757056

Special Sessions - Emerging Scope and Design Challenges for Approximate Computing: Optimizing Accuracy-PPA trade-offs and Beyond

Abstract

Publication series

Conference

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this