Grant: Alliance Grants
McMaster University
$25,000.00
Domain-Specific Compilers for Machine Learning in Cyber Physical Systems
FY 2023-24
Summary
Various goals such as efficiency, safety, resilience, and adaptability can be achieved by enabling the interaction and coordination of computational, physical, and human elements in cyber-physical systems (CPS). However, machine learning models that show promising results in several CPS goals often face real-time constraints due to the limited memory and computing resources in controllers used in CPS. Tensor algebra kernels, such as matrix multiplication, are required to execute Machine Learning (ML) models and optimization algorithms, stored as tensors, that are both memory and computation intensive. While many accelerators are used in high-performance servers to speed up ML workloads, industrial controllers often used in CPS do not have those features and usually rely on a single-core CPU.Moreover, the memory size in CPS processors ranges from 2KB to 512MB, and the processor is single-core with an instruction issue rate lower than GHz, which makes it much more restrictive than server processors. This project aims to develop automated software design methods and programming languages to enable machine learning and optimization algorithms for CPS. The new design from this project achieves various objectives such as real-time performance, memory and power efficiency, and resilience across different types of processors and controllers used in CPS.
Program
Alliance Grants
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
586319-2023
Name nom
Cheshmi, Kazem K
Department département
Electrical and Computer Engineering
Organization id
27
Institution établissement
McMaster University
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2023
Award amount
25000
Program id
ALLRP
Program name en
Alliance Grants
Program name fr
Subventions Alliance
Group en
RESEARCH PARTNERSHIPS
Group fr
PARTENARIATS DE RECHERCHE
Committee code
96
Committee name en
RPP Internal Decision Cttee
Committee name fr
Cté de décision interne - PPR
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2804
Research subject group en
Artificial intelligence
Research subject group fr
Expert systems
Research subject en
Intelligence artificielle
Research subject fr
Systèmes experts
Application title
Domain-Specific Compilers for Machine Learning in Cyber Physical Systems
Keyword
Compilers; Cyber-Physical Systems; Internt of Things; Machine Learning; Performance Optimization; Portable Software Design
Application summary
Various goals such as efficiency, safety, resilience, and adaptability can be achieved by enabling the interaction and coordination of computational, physical, and human elements in cyber-physical systems (CPS). However, machine learning models that show promising results in several CPS goals often face real-time constraints due to the limited memory and computing resources in controllers used in CPS. Tensor algebra kernels, such as matrix multiplication, are required to execute Machine Learning (ML) models and optimization algorithms, stored as tensors, that are both memory and computation intensive. While many accelerators are used in high-performance servers to speed up ML workloads, industrial controllers often used in CPS do not have those features and usually rely on a single-core CPU.Moreover, the memory size in CPS processors ranges from 2KB to 512MB, and the processor is single-core with an instruction issue rate lower than GHz, which makes it much more restrictive than server processors. This project aims to develop automated software design methods and programming languages to enable machine learning and optimization algorithms for CPS. The new design from this project achieves various objectives such as real-time performance, memory and power efficiency, and resilience across different types of processors and controllers used in CPS.