Reprogrammable Electro-Optic Nonlinear Activation Functions for Optical Neural Networks
نویسندگان
چکیده
منابع مشابه
Optical networks - the electro-optic reality
A general overview of the current status and future trends in optical networking is given. Special attention is given to the coexistence of optical and electronic technologies in telecommunications networks. After reviewing the advantages of both technologies, their use in different network areas is discussed. A critical survey of current and new transport technologies in optical core and metro...
متن کاملAerosol Deposition on Transparent Electro-Optic Films for Optical Modulators
Complex thin oxide films with electro-optic (EO) properties are promising for use in advanced optical devices because of their large EO effect. We developed a method of aerosol deposition (AD) for fabricating EO films. The mechanism for AD is based on the solidification by impact of submicron particles onto a substrate. Since particles in AD films preserve their crystalline structure during the...
متن کاملSilica/electro-optic Polymer Optical Modulator for Mmw Receiving (preprint)
A silica/electro-optic (EO) polymer phase modulator is proposed for millimeter-wave (MMW) radiation receiver with the use of a bowtie antenna. Waveguide design optimization is presented for a waveguide with an EO polymer core and silica/solgel cladding. Electrode effects on the insertion loss and poling efficiency are also analyzed and conditions for low-loss and high poling efficiency establis...
متن کاملStochastic Neural Networks with Monotonic Activation Functions
We propose a Laplace approximation that creates a stochastic unit from any smooth monotonic activation function, using only Gaussian noise. This paper investigates the application of this stochastic approximation in training a family of Restricted Boltzmann Machines (RBM) that are closely linked to Bregman divergences. This family, that we call exponential family RBM (Exp-RBM), is a subset of t...
متن کاملDeep Neural Networks with Multistate Activation Functions
We propose multistate activation functions (MSAFs) for deep neural networks (DNNs). These MSAFs are new kinds of activation functions which are capable of representing more than two states, including the N-order MSAFs and the symmetrical MSAF. DNNs with these MSAFs can be trained via conventional Stochastic Gradient Descent (SGD) as well as mean-normalised SGD. We also discuss how these MSAFs p...
متن کاملذخیره 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: IEEE Journal of Selected Topics in Quantum Electronics
سال: 2020
ISSN: 1077-260X,1558-4542
DOI: 10.1109/jstqe.2019.2930455