Ntroducing a Dversarial D Ropout in G Enera - Tive M Ulti - a Dversarial N Etworks

نویسندگان

  • Gonçalo Mordido
  • Haojin Yang
  • Christoph Meinel
چکیده

We propose to extend the original generative adversarial networks (GANs) framework to multiple discriminators and omit, or dropout, the feedback of each discriminator with same probability at the end of each batch. Our approach forces the generator to not rely on a given discriminator to learn how to produce realistic looking samples, but, instead, on a dynamic ensemble of adversaries. This promotes variety of the generated samples, leading to a richer generator less prone to mode collapsing. We show preliminary results on MNIST and Fashion-MNIST that sustain our claims.

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

Iclr 2018 S Pectral N Ormalization for G Enerative a Dversarial N Etworks

One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. Our new normalization technique is computationally light and easy to incorporate into existing implementations. We tested the efficacy of spectral n...

متن کامل

A Dversarial R Obustness of P Runed N Eural N Etworks

Deep neural network pruning forms a compressed network by discarding “unimportant” weights or filters. Standard evaluation metrics have shown their remarkable speedup and prediction accuracy in test time, but their adversarial robustness remains unexplored even though it is an important security feature in deployment. We study the robustness of pruned neural networks under adversarial attacks. ...

متن کامل

Verschränkung von Perspektiven durch Aushandlung

To foster c olla b ora tive lea rning a system sh ou ld su pport perspec tiva l view ing a nd neg otia tion in a na tu ra l w a y. W eb G u id e su pports tea m s of stu d ents c ond u c ting resea rc h on th e web ; it lets th em sh a re b ookm a rks, qu eries, notes a nd su m m a ries th a t a rise in th e resea rc h . Th e sh a red inform a tion spa c e is viewed th rou g h va riou s perspec...

متن کامل

A Dversarial a Utoencoders for N Ovelty D E - Tection

In this paper, we address the problem of novelty detection, i.e recognizing at test time if a data item comes from the training data distribution or not. We focus on Adversarial autoencoders (AAE) that have the advantage to explicitly control the distribution of the known data in the feature space. We show that when they are trained in a (semi-)supervised way, they provide consistent novelty de...

متن کامل

Nergy - Based G Enerative a Dversarial N Et - Works

We introduce the “Energy-based Generative Adversarial Network” model (EBGAN) which views the discriminator as an energy function that attributes low energies to the regions near the data manifold and higher energies to other regions. Similar to the probabilistic GANs, a generator is seen as being trained to produce contrastive samples with minimal energies, while the discriminator is trained to...

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 2018