Filters








5,531 Hits in 5.0 sec

Calibrating Energy-based Generative Adversarial Networks [article]

Zihang Dai, Amjad Almahairi, Philip Bachman, Eduard Hovy, Aaron Courville
2017 arXiv   pre-print
In this paper, we propose to equip Generative Adversarial Networks with the ability to produce direct energy estimates for samples.Specifically, we propose a flexible adversarial training framework, and  ...  Empirically, the experiment results closely match our theoretical analysis, verifying the discriminator is able to recover the energy of data distribution.  ...  INTRODUCTION Generative Adversarial Networks (GANs) (Goodfellow et al., 2014) represent an important milestone on the path towards more effective generative models.  ... 
arXiv:1702.01691v2 fatcat:piljftrdufbixcvwbzhplha6re

Towards Understanding the Generative Capability of Adversarially Robust Classifiers [article]

Yao Zhu, Jiacheng Ma, Jiacheng Sun, Zewei Chen, Rongxin Jiang, Zhenguo Li
2021 arXiv   pre-print
Based on our new understanding, we further propose a better adversarial training method, Joint Energy Adversarial Training (JEAT), which can generate high-quality images and achieve new state-of-the-art  ...  We reformulate adversarial example generation, adversarial training, and image generation in terms of an energy function.  ...  We also propose Preliminary Joint Energy Adversarial Training (PreJEAT) which just trains model with adversarial examples found by Eq. (18) and use cross-entropy loss.  ... 
arXiv:2108.09093v2 fatcat:zrnozsceyfhkvjfzd2ipzny7hi

Likelihood-Free Inference with Generative Neural Networks via Scoring Rule Minimization [article]

Lorenzo Pacchiardi, Ritabrata Dutta
2022 arXiv   pre-print
However, generative networks only allow sampling from the parametrized distribution; for this reason, Ramesh et al. [2022] follows the common solution of adversarial training, where the generative network  ...  Here, we propose to approximate the posterior with generative networks trained by Scoring Rule minimization, an overlooked adversarial-free method enabling smooth training and better uncertainty quantification  ...  For this reason, maximum likelihood estimation of neural network weights is impossible and people use training methods based on generating samples from the generative network.  ... 
arXiv:2205.15784v1 fatcat:j55z4maqfvcrrcxa44s636lbzu

Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One [article]

Will Grathwohl, Kuan-Chieh Wang, Jörn-Henrik Jacobsen, David Duvenaud, Mohammad Norouzi, Kevin Swersky
2020 arXiv   pre-print
We demonstrate that energy based training of the joint distribution improves calibration, robustness, andout-of-distribution detection while also enabling our models to generate samplesrivaling the quality  ...  We propose to reinterpret a standard discriminative classifier of p(y|x) as an energy based model for the joint distribution p(x,y).  ...  We would also like to thank Jeremy Cohen for his useful feedback which greatly strengthened our adversarial robustness results.  ... 
arXiv:1912.03263v3 fatcat:6asqbqn3ffajrgohcwuqnaa3li

Energy-Efficient Implementation of Generative Adversarial Networks on Passive RRAM Crossbar Arrays [article]

Siddharth Satyam, Honey Nikam, Shubham Sahay
2022 arXiv   pre-print
Moreover, the frequent data transfer between the discriminative and generative networks during training significantly degrades the efficacy of the von-Neumann GAN accelerators such as those based on GPU  ...  However, the adversarial (competitive) co-training of the discriminative and generative networks in GAN makes them computationally intensive and hinders their deployment on the resource-constrained IoT  ...  Generative Adversarial Networks (GANs) Fully-connected GANs rely on an adversarial training process that involves a zero-sum game between two multi-layer perceptrons: the generator and the discriminator  ... 
arXiv:2111.14484v2 fatcat:44j5uwufb5fbtdp5a4sdvkdzve

MRI Reconstruction Using Deep Energy-Based Model [article]

Yu Guan, Zongjiang Tu, Shanshan Wang, Qiegen Liu, Yuhao Wang, Dong Liang
2021 arXiv   pre-print
Purpose: Although recent deep energy-based generative models (EBMs) have shown encouraging results in many image generation tasks, how to take advantage of the self-adversarial cogitation in deep EBMs  ...  Leveraging this, a novel regularization strategy is introduced in this article which takes advantage of self-adversarial cogitation of the deep energy-based model.  ...  Yilun Du for the previous work on the deep energy-based models and codes [25] that are very helpful in this paper. This work was in part supported by Na-  ... 
arXiv:2109.03237v2 fatcat:s23takonkfaxpdr4lvlpizinw4

Noise Sensitivity-Based Energy Efficient and Robust Adversary Detection in Neural Networks [article]

Rachel Sterneck, Abhishek Moitra, Priyadarshini Panda
2021 arXiv   pre-print
Based on prior works on detecting adversaries, we propose a structured methodology of augmenting a deep neural network (DNN) with a detector subnetwork.  ...  Furthermore, we validate the energy efficiency of our proposed adversarial detection methodology through an extensive energy analysis on various hardware scalable CMOS accelerator platforms.  ...  We propose ANS-based detectors as a robust and efficient method for preventing adversarial attacks, and we encourage future research to continue analyzing adversarial examples from a structural perspective  ... 
arXiv:2101.01543v1 fatcat:k7uy7tr6grfcrp3ua4nlf54coa

Entropic Out-of-Distribution Detection [article]

David Macêdo, Tsang Ing Ren, Cleber Zanchettin, Adriano L. I. Oliveira, Teresa Ludermir
2021 arXiv   pre-print
Moreover, our experiments show that training neural networks with IsoMax loss significantly improves their OOD detection performance.  ...  ., classification accuracy drop, slower energy-inefficient inferences).  ...  Solutions based on uncertainty (or confidence) estimation (or calibration) [29] - [33] usually present additional complexity, slow and energy-inefficient inferences [15] , and OOD detection performance  ... 
arXiv:1908.05569v13 fatcat:erczjcwhjzdnlgn3bksxwjvipi

Towards the Increase in Granularity for the Main Hadronic ATLAS Calorimeter: Exploiting Deep Learning Methods

Philipp Gaspar
2020 Journal of Physics, Conference Series  
This paper presents the latest results from using a Generative Adversarial Network (GAN) to generate synthetic images, which simulate real images formed in the MA-PMT.  ...  Generative Adversarial Networks (GANs) The generative model will be developed for increasing the statistics. The GAN approach is based in two deep neural networks that compete with each other.  ...  First, increasing the statistics with simulation generated by an adversarial model, and then performing a binary classification algorithm based on deep neural networks.  ... 
doi:10.1088/1742-6596/1525/1/012094 fatcat:tiqdghu64jdejhhgtmdj4bjaza

Enabling Dark Energy Science with Deep Generative Models of Galaxy Images [article]

Siamak Ravanbakhsh and Francois Lanusse and Rachel Mandelbaum and Jeff Schneider and Barnabas Poczos
2016 arXiv   pre-print
In particular we consider variations on conditional variational autoencoder and introduce a new adversarial objective for training of conditional generative networks.  ...  Our results suggest a reliable alternative to the acquisition of expensive high quality observations for generating the calibration data needed by the next generation of cosmological surveys.  ...  autoencoder (C-VAE, scheme I) and our variation on conditional generative adversarial network (C-GAN).  ... 
arXiv:1609.05796v2 fatcat:bk5kc5f4wbe7pbfhqataeggzfm

Two Generator Game: Learning to Sample via Linear Goodness-of-Fit Test

Lizhong Ding, Mengyang Yu, Li Liu, Fan Zhu, Yong Liu, Yu Li, Ling Shao
2019 Neural Information Processing Systems  
network is trained by minimizing a GOF test statistic between the energy-based distribution and the generated data, such that the underlying distribution of the generated data is close to the energy-based  ...  DEAN can be interpreted as a GOF game between two generative networks, where one explicit generative network learns an energy-based distribution that fits the real data, and the other implicit generative  ...  Deep Energy Adversarial Network In this section, we present the deep energy adversarial network (DEAN).  ... 
dblp:conf/nips/DingYL0LLS19 fatcat:aghnut5zljfhvk75mwauetpyhe

Front Matter: Volume 11595

Hilde Bosmans, Wei Zhao, Lifeng Yu
2021 Medical Imaging 2021: Physics of Medical Imaging  
adversarial networks and radiomics supervision for lung lesion synthesis (Robert F.  ...  image-based convolutional neural network for 4D-CBCT reconstructions enhancement 11595 1U Deep learning in image reconstruction: vulnerability under adversarial attacks and potential defense strategies  ...  generative adversarial network (Res-CycleGAN) 11595 42 Investigation of the efficacy of a data-driven CT artifact correction scheme for sparse and truncated projection data for intracranial hemorrhage  ... 
doi:10.1117/12.2595450 fatcat:dji2t6dpdfantjoyxtpdik6yvu

Generative Max-Mahalanobis Classifiers for Image Classification, Generation and More [article]

Xiulong Yang, Hui Ye, Yang Ye, Xiang Li, Shihao Ji
2021 arXiv   pre-print
Joint Energy-based Model (JEM) of Grathwohl et al. shows that a standard softmax classifier can be reinterpreted as an energy-based model (EBM) for the joint distribution p(x,y); the resulting model can  ...  be optimized to improve calibration, robustness, and out-of-distribution detection, while generating samples rivaling the quality of recent GAN-based approaches.  ...  Background and Related Work Energy-based Models Energy-based models (EBMs) [21] define an energy function that assigns low energy values to samples drawn from data distribution and high values otherwise  ... 
arXiv:2101.00122v4 fatcat:kpahcnlfljcercpuf6hkxfcnri

Closing the sim-to-real gap in guided wave damage detection with adversarial training of variational auto-encoders [article]

Ishan D. Khurjekar, Joel B. Harley
2022 arXiv   pre-print
In this work, we counter this challenge by training an ensemble of variational autoencoders only on simulation data with a wave physics-guided adversarial component.  ...  These include strategies based on likelihood models [10] , autoregressive networks, [11] , adversarial networks [12] , and variational autoencoders (VAE) [13] , among others.  ...  The calibration bank is considered as an extension of the validation set. For a particular test signal, we choose the calibration signal that minimizes residual energy.  ... 
arXiv:2202.00570v1 fatcat:eeh6xj2cknhe3fvjnvp7fpbwtm

Blackhole Attack Detection Techniques in WSN: A Review

Mandeep Thakur, Amninder Kaur
2017 International Journal of Advanced Research in Computer Science and Software Engineering  
The wireless sensor network is the decentralized type of network, the size of the sensor networks is very small due to which battery power of the nodes is limited.  ...  Due to self configuring nature of the wireless sensor networks, various type security attacks are possible in the network. The security attacks are broadly classified into active and passive attacks.  ...  In blackhole attack, an adversary captures and re-programs a set of nodes in the network to block the packets they receive instead of forwarding them towards the base station.  ... 
doi:10.23956/ijarcsse/v7i4/0225 fatcat:qymfpu43vre3rda73dhmefy6di
« Previous Showing results 1 — 15 out of 5,531 results