0% Complete
صفحه اصلی
/
ششمین کنفرانس بین المللی میکروالکترونیک ایران
Dynamic Power Control in a Hardware Neural Network with Error-Configurable MAC Units
نویسندگان :
Maedeh Ghaderi
1
Arvin Delavari
2
Faraz Ghoreishy
3
Sattar Mirzakuchaki
4
1- Iran university of science and technology
2- Iran university of science and technology
3- Iran university of science and technology
4- Iran university of science and technology
کلمات کلیدی :
Neural networks،hardware accelerators،approximate computing،low-power design،very large-scale integration،image classification
چکیده :
Multi-Layer Perceptrons (MLP) are powerful tools for representing complex, non-linear relationships, making them essential for diverse machine learning and AI applications. Efficient hardware implementation of MLPs can be achieved through many hardware and architectural design techniques. These networks excel at predictive modeling and classification tasks like image classification, making them a popular choice. Approximate computing techniques are increasingly used to optimize critical path delay, area, power, and overall hardware efficiency in high-performance computing systems through controlled error and related trade-offs. This study proposes a hardware MLP neural network implemented in 45nm CMOS technology, in which MAC units of the neurons incorporate error and power controllable approximate multipliers for classification of the MNIST dataset. The optimized network consists of 10 neurons within the hidden layers, occupying 0.026mm² of area, with 5.55mW at 100MHz frequency in accurate mode and 4.81mW in lowest accuracy mode. The experiments indicate that the proposed design achieves a maximum rate of 13.33% decrease overall and 24.78% in each neuron’s power consumption with only a 0.92% decrease in accuracy in comparison with accurate circuit.
لیست مقالات
لیست مقالات بایگانی شده
Dynamic Power Control in a Hardware Neural Network with Error-Configurable MAC Units
Maedeh Ghaderi - Arvin Delavari - Faraz Ghoreishy - Sattar Mirzakuchaki
A Low-Power Differential Ring VCO Using An Active Inductor For Wireless Applications
Mahdi Alijani - Mohammadmahdi Javanmardi - Adib Abrishamifar
Comparison between Hardware/Software Co-design of RiscV Vector and Scalar Implementation of Deep Neural Networks
Seyed Kian Mousavikia - Morteza Mousazadeh
یک مبدل آنالوگ به دیجیتال مبتنی بر مدولاتور سیگما دلتا برای کاربردهای مهندسی - پزشکی با ENOB = 13. 2 bits ، پهنای باند 10 kHz و توان مصرفی 16.9 µ W
علی صداقت - حسین پاک نیت - نوید یثربی
Using GDI Structure in Hardware Implementation of Convolution Operation in Deep Neural Networks
Maedeh Kadkhodaie - Sayed Masoud Sayedi
A MEMS Resonant Pressure Sensore Based on 2D Graphene Material
Amir Noroolahi - Farshad Babazadeh
Design of a Calibration Circuit for Adaptive Phase-Locked Loop in the 5GHz Range Using CMOS 180nm Technology
Reza MirAlvandi - Mahdi Ehsanian
A Low-Power Fully Differential LC Oscillator with Phase Noise Reduction for LTE Applications
Yeganeh Moradzadeh Rezaei - SIROUS TOOFAN - Jafar Sobhi
Design of Electrical Stimulation Circuit in 180 nm/1.8 V Standard CMOS Process
Askandar Nikzad - Mohammad Yavari - Amir Kashi
First Principles Study of Optical and Electrical Properties for Mixed-halide 2D BA2PbBr4-xClx (x=0, 2, and 4) as an Active Layer of Perovskite Light Emitting Diode
ُSamad Shokouhi - Seyedeh Bita Saadatmand - Vahid Ahmadi
بیشتر
ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 41.5.5