no code implementations • 5 Mar 2024 • Mahdi Taheri, Masoud Daneshtalab, Jaan Raik, Maksim Jenihhin, Salvatore Pappalardo, Paul Jimenez, Bastien Deveautour, Alberto Bosio
Systolic array has emerged as a prominent architecture for Deep Neural Network (DNN) hardware accelerators, providing high-throughput and low-latency performance essential for deploying DNNs across diverse applications.
no code implementations • 5 Mar 2024 • Mahdi Taheri, Natalia Cherezova, Samira Nazari, Ahsan Rafiq, Ali Azarpeyvand, Tara Ghasempouri, Masoud Daneshtalab, Jaan Raik, Maksim Jenihhin
In this paper, we propose an architecture of a novel adaptive fault-tolerant approximate multiplier tailored for ASIC-based DNN accelerators.
no code implementations • 17 Jan 2024 • Mahdi Taheri, Natalia Cherezova, Mohammad Saeed Ansari, Maksim Jenihhin, Ali Mahani, Masoud Daneshtalab, Jaan Raik
The stringent requirements for the Deep Neural Networks (DNNs) accelerator's reliability stand along with the need for reducing the computational burden on the hardware platforms, i. e. reducing the energy consumption and execution time as well as increasing the efficiency of DNN accelerators.
no code implementations • 16 Jun 2023 • Mohammad Hasan Ahmadilivani, Mahdi Taheri, Jaan Raik, Masoud Daneshtalab, Maksim Jenihhin
Thereafter, a novel method for splitting the critical neurons is proposed that enables the design of a Lightweight Correction Unit (LCU) in the accelerator without redesigning its computational part.
no code implementations • 31 May 2023 • Mahdi Taheri, Saeideh Sheikhpour, Ali Mahani, Maksim Jenihhin
We introduce a novel logic style with self-checking capability to enhance hardware reliability at logic level.
no code implementations • 31 May 2023 • Mahdi Taheri, Mohammad Hasan Ahmadilivani, Maksim Jenihhin, Masoud Daneshtalab, Jaan Raik
Nowadays, the extensive exploitation of Deep Neural Networks (DNNs) in safety-critical applications raises new reliability concerns.
no code implementations • 31 May 2023 • Mohammad Hasan Ahmadilivani, Mario Barbareschi, Salvatore Barone, Alberto Bosio, Masoud Daneshtalab, Salvatore Della Torca, Gabriele Gavarini, Maksim Jenihhin, Jaan Raik, Annachiara Ruospo, Ernesto Sanchez, Mahdi Taheri
We propose to use approximate (AxC) arithmetic circuits to agilely emulate errors in hardware without performing fault injection on the DNN.
no code implementations • 9 May 2023 • Mohammad Hasan Ahmadilivani, Mahdi Taheri, Jaan Raik, Masoud Daneshtalab, Maksim Jenihhin
Through this SLR, three kinds of methods for reliability assessment of DNNs are identified including Fault Injection (FI), Analytical, and Hybrid methods.
no code implementations • 26 Apr 2023 • Mahtab Taheri, Mahdi Taheri, Amirhossein Hadjahmadi
In recent years, a person's age has been studied as one of the important features of the face.
no code implementations • 15 Mar 2023 • Ahmad Rezaei, Mahdi Taheri, Ali Mahani, Sebastian Magierowski
In this paper, we discuss several shortcomings of prior base callers in the case of time-critical applications, privacy-aware design, and the problem of catastrophic forgetting.
no code implementations • 14 Mar 2023 • Mahdi Taheri, Mohammad Riazati, Mohammad Hasan Ahmadilivani, Maksim Jenihhin, Masoud Daneshtalab, Jaan Raik, Mikael Sjodin, Bjorn Lisper
The framework enables selective approximation of reliability-critical DNNs, providing a set of Pareto-optimal DNN implementation design space points for the target resource utilization requirements.
no code implementations • 13 Mar 2023 • Mohammad Hasan Ahmadilivani, Mahdi Taheri, Jaan Raik, Masoud Daneshtalab, Maksim Jenihhin
In this work, we propose a novel accurate, fine-grain, metric-oriented, and accelerator-agnostic method called DeepVigor that provides vulnerability value ranges for DNN neurons' outputs.