Methods for reliability assessment and enhancement of deep neural networks hardware accelerators = Süvanärvivõrkude riistvara kiirendite töökindluse hindamine ja täiustamine
author
supervisor
statement of authorship
Mahdi Taheri ; [supervisors: Maksim Jenihhin, Masoud Daneshtalab ; Tallinn University of Technology, School of Information Technologies, Department of Computer Systems]
type of dissertation
doktoritöö
university/scientific institution
Tallinna Tehnikaülikool
location of publication
Tallinn
publisher
year of publication
pages
288 p. : ill
series
Tallinn University of Technology. Doctoral thesis = Tallinna Tehnikaülikool. Doktoritöö ; 4/2025
subject term
subject of form
ISSN
2585-6898
2585–6901 (PDF)
ISBN
978-9916-80-251-9 (PDF)
978-9916-80-250-2
notes
Autori publikatsioonide nimekiri lehekülgedel 8-9
Bibliograafia lehekülgedel 110-124
Ilmunud ka võrguressursina
Kokkuvõte eesti keeles
Kättesaadav ka võrguteavikuna
Autori CV inglise ja eesti keeles, lk. 287-288
Thesis (Ph.D.in Information and Communication Technologies) : Tallinn University of Technology, 2025
url
Open Access
Open Access
scientific publication
teaduspublikatsioon
classifier
TalTech department
language
inglise
- APPRAISER : DNN fault resilience analysis employing approximation errors
- DeepAxe : a framework for exploration of approximation and reliability trade-offs in DNN accelerators
- Special session : approximation and fault resiliency of DNN accelerators
- A systematic literature review on hardware reliability assessment methods for deep neural networks
- AdAM: adaptive fault-tolerant approximate multiplier for edge DNN accelerators
- Special session: reliability assessment recipes for DNN accelerators
- Keynote: cost-efficient reliability for Edge-AI chips
- Enhancing fault resilience of QNNs by selective neuron splitting
- DeepVigor: VulnerabIlity Value RanGes and FactORs for DNNs’ Reliability Assessment
- Exploration of Activation Fault Reliability in Quantized Systolic Array-Based DNN Accelerators
- AdAM: Adaptive Approximate Multiplier for Fault Tolerance in DNN Accelerators
- SAFFIRA: a Framework for Assessing the Reliability of Systolic-Array-Based DNN Accelerators
- FORTUNE: A Negative Memory Overhead Hardware-Agnostic Fault TOleRance TechniqUe in DNNs
- Heterogeneous Approximation of DNN HW Accelerators based on Channels Vulnerability
Taheri, M. Methods for reliability assessment and enhancement of deep neural networks hardware accelerators = Süvanärvivõrkude riistvara kiirendite töökindluse hindamine ja täiustamine. Tallinn : TalTech Press, 2025. 288 p. : ill. (Tallinn University of Technology. Doctoral thesis = Tallinna Tehnikaülikool. Doktoritöö ; 4/2025).