Efficient neural network accelerators with optical computing and communication

نویسندگان

چکیده

Conventional electronic Artificial Neural Networks (ANNs) accelerators focus on architecture design and numerical computation optimization to improve the training efficiency. However, these approaches have recently encountered bottlenecks in terms of energy efficiency computing performance, which leads an increase interest photonic accelerator. Photonic architectures with low consumption, high transmission speed bandwidth been considered as important role for generation architectures. In this paper, provide a better understanding optical technology used ANN acceleration, we present comprehensive review efficient communication accelerators. The related devices are investigated application ANNs classification existing solutions is proposed that categorized into acceleration according effects Moreover, discuss challenges neural network highlight most promising future research opportunities field.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Neural Network with Binary Activations for Efficient Neuromorphic Computing

In this paper, we proposed techniques to train and deploy a multi-layer neural network using softmax loss for binary activations to best capitalize on its energy efficiency. These techniques include using the gradient of tanh function to approximate gradient of discrete binary threshold function during the backpropagation of training, and using a stochastic multi-sampling approach to convert hi...

متن کامل

the relationship of wtc with communication apprehension and self-perceived communication competene in english and persian context

بیشتر تحقیقات پیشین در زمینه تمایل به برقراری ارتباط به رابطه آن با عوامل فردی چون سن، جنس، نوع شخصیت و... صورت گرفته است. در صورتی که مطالعات کمتری به بررسی رابطه تمایل به برقراری ارتباط زبان آموزان فارسی زبان با ترس از برقراری ارتباط و توانش خود ادراکانه آنها در برقراری ارتباط در محیط فارسی و انگلیسی انجام شده است. بر اساس نظریه الیس (2008) تمایل به برقراری ارتباط جایگاه مهمی در زمینه آموزش م...

15 صفحه اول

High-Performance Computing with Accelerators

T his issue of CiSE is based on work presented at the US National Science Foundation workshop, “Path to Petascale: Adapting Geo/Chem/Astro Applications for Accelerators and Accelerator Clusters,” held at the US National Center for Supercomputing Appli cations (NCSA) in early 2009. The workshop was designed to raise awareness about the emergence of accelerator-based high-performance computing (H...

متن کامل

ThUnderVolt: Enabling Aggressive Voltage Underscaling and Timing Error Resilience for Energy Efficient Deep Neural Network Accelerators

Hardware accelerators are being increasingly deployed to boost the performance and energy efficiency of deep neural network (DNN) inference. In this paper we propose Thundervolt, a new framework that enables aggressive voltage underscaling of high-performance DNN accelerators without compromising classification accuracy even in the presence of high timing error rates. Using post-synthesis timin...

متن کامل

scour modeling piles of kambuzia industrial city bridge using hec-ras and artificial neural network

today, scouring is one of the important topics in the river and coastal engineering so that the most destruction in the bridges is occurred due to this phenomenon. whereas the bridges are assumed as the most important connecting structures in the communications roads in the country and their importance is doubled while floodwater, thus exact design and maintenance thereof is very crucial. f...

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Computer Science and Information Systems

سال: 2023

ISSN: ['1820-0214', '2406-1018']

DOI: https://doi.org/10.2298/csis220131066x