Bayesian GAN

نویسندگان

  • Yunus Saatci
  • Andrew Wilson
چکیده

Generative adversarial networks (GANs) can implicitly learn rich distributions over images, audio, and data which are hard to model with an explicit likelihood. We present a practical Bayesian formulation for unsupervised and semi-supervised learning with GANs. Within this framework, we use stochastic gradient Hamiltonian Monte Carlo to marginalize the weights of the generator and discriminator networks. The resulting approach is straightforward and obtains good performance without any standard interventions such as feature matching or mini-batch discrimination. By exploring an expressive posterior over the parameters of the generator, the Bayesian GAN avoids mode-collapse, produces interpretable and diverse candidate samples, and provides state-of-the-art quantitative results for semi-supervised learning on benchmarks including SVHN, CelebA, and CIFAR-10, outperforming DCGAN, Wasserstein GANs, and DCGAN ensembles.

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

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Bayesian Renewables Scenario Generation via Deep Generative Networks

We present a method to generate renewable scenarios using Bayesian probabilities by implementing the Bayesian generative adversarial network (Bayesian GAN), which is a variant of generative adversarial networks based on two interconnected deep neural networks. By using a Bayesian formulation, generators can be constructed and trained to produce scenarios that capture different salient modes in ...

متن کامل

Non-linear modeling, analysis, design and simulation of a solid state power amplifier based on GaN technology for Ku band microwave application

A new non-linear method for design and analysis of solid state power amplifiers is presented and applied to an aluminum gallium nitride, gallium nitride (AlGaN-GaN) high electron-mobility transistor (HEMTs) on silicon-carbide (SiC) substrate for Ku band (12.4 13.6 GHz) applications. With combining output power of 8 transistors, maximum output power of 46.3 dBm (42.6 W), PAE of 43% and linear ga...

متن کامل

وابستگی انرژی گذارهای اپتیکی در نانوساختارهای چاههای کوانتومی GaN/AlGaN به پهنای سد و چاه کوانتومی

Internal polarizations field which take place in quantum structures of group-III nitrides have an important consequence on their optical properties. Optical properties of wurtzite AlGaN/GaN quantum well (QW) structures grown by MBE and MOCVD on c-plane sapphire substrates have been investigated by means of photoluminescence (PL) and time resolved photoluminescence (TRPL) at low-temperature. PL ...

متن کامل

Numerical Modeling of Electronic and Electrical Characteristics of 0.3 0.7 Al Ga N / GaN Multiple Quantum Well Solar Cells

The present study was conducted to investigate current density of0.3 0.7 Al Ga N/ GaN multiple quantum well solar cell (MQWSC) under hydrostaticpressure. The effects of hydrostatic pressure were taken into account to measureparameters of 0.3 0.7 Al Ga N/ GaN MQWSC, such as interband transition energy, electronholewave functions, absorption coefficient, and dielectric con...

متن کامل

ذخیره در منابع من


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

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2017