Title:
Energy-aware DNN Quantization for Processing-In-Memory Architecture
Energy-aware DNN Quantization for Processing-In-Memory Architecture
dc.contributor.advisor | Mukhopadhyay, Saibal | |
dc.contributor.advisor | Yu, Shimeng | |
dc.contributor.advisor | Krishna, Tushar | |
dc.contributor.author | Kang, Beomseok | |
dc.contributor.department | Electrical and Computer Engineering | |
dc.date.accessioned | 2022-08-25T13:31:40Z | |
dc.date.available | 2022-08-25T13:31:40Z | |
dc.date.created | 2022-08 | |
dc.date.issued | 2022-05-13 | |
dc.date.submitted | August 2022 | |
dc.date.updated | 2022-08-25T13:31:41Z | |
dc.description.abstract | With increasing computational cost of deep neural network (DNN), many efforts to develop energy-efficient intelligent system have been proposed from dedicated hardware platforms to model compression algorithms. Recently, hardware-aware quantization algorithms have shown further improvement in the energy efficiency of DNN by considering hardware architectures and algorithms together. In this work, a genetic algorithm-based energy-aware DNN quantization framework for Processing-In-Memory (PIM) architectures, named EGQ, is presented. The key contribution of the research is to design a fitness function that can reduce the number of analog-to-digital converter (ADC) access, which is one of the main energy overhead in PIM. EGQ automatically optimizes layer-wise weight and activation bitwidth with negligible accuracy loss while considering the dynamic energy in PIM. The research demonstrates the effectiveness of EGQ on several DNN models VGG-19, ResNet-18, ResNet-50, MobileNet-V2, and SqueezeNet. Also, the area, dynamic energy, and energy efficiency in the compressed models with various memory technologies are analyzed. EGQ shows 15%-103% higher energy efficiency with 2% accuracy loss than other PIM-aware quantization algorithms. | |
dc.description.degree | M.S. | |
dc.format.mimetype | application/pdf | |
dc.identifier.uri | http://hdl.handle.net/1853/67193 | |
dc.language.iso | en_US | |
dc.publisher | Georgia Institute of Technology | |
dc.subject | Quantization | |
dc.subject | Deep Neural Network | |
dc.subject | Genetic Algorithm | |
dc.subject | Processing-In-Memory | |
dc.title | Energy-aware DNN Quantization for Processing-In-Memory Architecture | |
dc.type | Text | |
dc.type.genre | Thesis | |
dspace.entity.type | Publication | |
local.contributor.advisor | Mukhopadhyay, Saibal | |
local.contributor.advisor | Yu, Shimeng | |
local.contributor.advisor | Krishna, Tushar | |
local.contributor.corporatename | School of Electrical and Computer Engineering | |
local.contributor.corporatename | College of Engineering | |
relation.isAdvisorOfPublication | 62df0580-589a-4599-98af-88783123945a | |
relation.isAdvisorOfPublication | 95da035d-d0bb-439b-b9f1-7a4ab7f18bab | |
relation.isAdvisorOfPublication | f80c3b14-cd42-456d-b440-addf20372fbc | |
relation.isOrgUnitOfPublication | 5b7adef2-447c-4270-b9fc-846bd76f80f2 | |
relation.isOrgUnitOfPublication | 7c022d60-21d5-497c-b552-95e489a06569 | |
thesis.degree.level | Masters |