Loading...
Search
Search in this resource
sort by
Send to Friend
 Digital Object List
 Bookmark
 Review List
  1. طراحی واحد پردازشگر نوری جهت انجام عملیات غیرخطی در شبکه‌های یادگیری عمیق
طراحی واحد پردازشگر نوری جهت انجام عملیات غیرخطی در شبکه‌های یادگیری عمیق
406 viewed

طراحی واحد پردازشگر نوری جهت انجام عملیات غیرخطی در شبکه‌های یادگیری عمیق

ابراهیمی دهقان پور، آیدا Ebrahimi Dehghanpour, Aida

Designing an Optical Processing Unit for Non-Linear Operations in Deep Neural Networks

Ebrahimi Dehghanpour, Aida | 2022

406 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54798 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Koohi, Somayyeh
  7. Abstract:
  8. Abstract: In this thesis, we tackled the problem of nonlinear activation function in optical artificial neural networks, and in particular in convolutional and recurrent neural networks. In the end, we propose an all-optical recurrent neural network in free-space optics for the first time. Artificial neural networks are a branch of artificial intelligence, which can be adopted to solve a wide variety of problems. While very powerful, these networks can be very power hungry and slow when it comes to solving very complicated problems. Optical versions of these networks bring the promise of solving both of these issues and provide a fast and power efficient platform for these networks. However, providing a nonlinear activation function which is a necessary part of artificial neural networks, is very challenging. In this regard, we proposed an all-optical programmable activation function by the means of two saturable absorbers and a programmable optical adder/subtractor. As a key advantage, the proposed activation function take various forms and be programmed to both act as a bounded and unbounded function and resemble many of the well-known activation functions (i.e. Sigmoid, Tanh, and ReLU). Thus, this activation function can be used in various neural networks ranging from LSTM to convolutional neural networks. Moreover, as this activation function is programmable, it can pave the way for an all-optical general processing unit
  9. Keywords:
  10. Optical Neural Networks ; Convolutional Neural Network ; Recurrent Neural Networks ; Parallel Processing ; Ultra Fast Processing ; Nonlinear Optical Unit

 Digital Object List

 Bookmark

No TOC