Filters








32,885 Hits in 14.4 sec

Domain Adversarial Neural Networks for Domain Generalization: When It Works and How to Improve [article]

Anthony Sicilia, Xingchen Zhao, Seong Jae Hwang
2022 arXiv   pre-print
In particular, we note the bound on target error given by Ben-David et al. (2010) and the well-known domain-aligning algorithm based on this work using Domain Adversarial Neural Networks (DANN) presented  ...  Our investigation suggests that the application of DANN to domain generalization may not be as straightforward as it seems.  ...  Domain Adversarial Neural Network (DANN) In this section, we cover the necessary background on Domain Adversarial Neural Networks (DANN).  ... 
arXiv:2102.03924v2 fatcat:bypoj64xanay3gl7yerxamutii

Domain Adaptation for Relation Extraction with Domain Adversarial Neural Network

Lisheng Fu, Thien Huu Nguyen, Bonan Min, Ralph Grishman
2017 International Joint Conference on Natural Language Processing  
Our method is a joint model consisting of a CNN-based relation classifier and a domain-adversarial classifier.  ...  Trained on a source domain, a relation extractor's performance degrades when applied to target domains other than the source.  ...  Domain Adversarial Neural Network How does domain adaptation work without any labeled examples for the target domain?  ... 
dblp:conf/ijcnlp/FuNMG17 fatcat:3ur6ziflarevfnv3gxq5w52yem

Domain Adversarial Neural Networks for Dysarthric Speech Recognition [article]

Dominika Woszczyk, Stavros Petridis, David Millard
2020 arXiv   pre-print
This work explores domain adversarial neural networks (DANN) for speaker-independent speech recognition on the UAS dataset of dysarthric speech.  ...  We also observe that when labelled dysarthric speech data is available DANN and MTL perform similarly, but when they are not DANN performs better than MTL.  ...  Adversarial training using domain adversarial training of neural network (DANN) has been successfully applied to unsupervised domain adaptation in computer vision [12] and ASR [13, 14] , making it an  ... 
arXiv:2010.03623v1 fatcat:l4e5nhsjarekngdtgce23fdmzm

Domain Adversarial Neural Networks for Dysarthric Speech Recognition

Dominika Woszczyk, Stavros Petridis, David Millard
2020 Interspeech 2020  
This work explores domain adversarial neural networks (DANN) for speaker-independent speech recognition on the UAS dataset of dysarthric speech.  ...  We also observe that when labelled dysarthric speech data is available DANN and MTL perform similarly, but when they are not DANN performs better than MTL.  ...  Adversarial training using domain adversarial training of neural network (DANN) has been successfully applied to unsupervised domain adaptation in computer vision [12] and ASR [13, 14] , making it an  ... 
doi:10.21437/interspeech.2020-2845 dblp:conf/interspeech/WoszczykPM20 fatcat:cp66uzgv5zejffdhbilc6wx4w4

Learning Multi-Domain Adversarial Neural Networks for Text Classification

Xiao Ding, Qiankun Shi, Bibo Cai, Ting Liu, Yanyan Zhao, Qiang Ye
2019 IEEE Access  
Deep neural networks have been applied to learn transferable features for adapting text classification models from a source domain to a target domain.  ...  baselines, and it is general enough to be applied to more scenarios.  ...  They are grateful to Wayne Xin Zhao for sharing annotated positive instances of Phone domain with us.  ... 
doi:10.1109/access.2019.2904858 fatcat:gxg7vmhmdbd7jbn7rptvovnzrq

Context-Dependent Domain Adversarial Neural Network for Multimodal Emotion Recognition

Zheng Lian, Jianhua Tao, Bin Liu, Jian Huang, Zhanlei Yang, Rongjun Li
2020 Interspeech 2020  
To address these difficulties, we focus on the domain adversarial neural networks (DANN) for emotion recognition. The primary task is to predict emotion labels.  ...  In this paper, we propose the context-dependent domain adversarial neural network for multimodal emotion recognition.  ...  Domain Adversarial Neural Networks DANN has two classifiers -the emotion classifier and the domain classifier.  ... 
doi:10.21437/interspeech.2020-1705 dblp:conf/interspeech/LianTL0YL20 fatcat:r6boq2jsa5dcpa5i434rfdny2m

Adversarial Weighting for Domain Adaptation in Regression [article]

Antoine de Mathelin, Guillaume Richard, Francois Deheeger, Mathilde Mougeot, Nicolas Vayatis
2021 arXiv   pre-print
To solve this problem, we develop an adversarial network algorithm which learns both the source weighting scheme and the task in one feed-forward gradient descent.  ...  We introduce a novel formulation of the optimization objective for domain adaptation which relies on a discrepancy distance characterizing the difference between domains according to a specific task and  ...  Adversarial domain adaptation Adversarial training of neural networks have been first used for domain adaptation with DANN [13] .  ... 
arXiv:2006.08251v4 fatcat:3mlm2jptlraapngjgmdp6cbz7e

Domain Adversarial Neural Networks for Large-Scale Land Cover Classification

Bejiga, Melgani, Beraldini
2019 Remote Sensing  
In this work, we propose an unsupervised DA technique called domain adversarial neural networks (DANNs), composed of a feature extractor, a class predictor, and domain classifier blocks, for large-scale  ...  Additionally, we also modify the domain classifier of a DANN to evaluate its suitability for multi-target domain adaptation problems.  ...  Acknowledgments: We would like to thank the U.S Geological Survey for providing Landsat 8 images used in this work. Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/rs11101153 fatcat:s6mqbnwj2zhvterors3yxx3gkq

Incorporating human and learned domain knowledge into training deep neural networks: A differentiable dose volume histogram and adversarial inspired framework for generating Pareto optimal dose distributions in radiation therapy [article]

Dan Nguyen, Rafe McBeth, Azar Sadeghnejad Barkousaraie, Gyanendra Bohara, Chenyang Shen, Xun Jia, Steve Jiang
2019 arXiv   pre-print
We propose a novel domain specific loss, which is a differentiable loss function based on the dose volume histogram, and combine it with an adversarial loss for the training of deep neural networks to  ...  Expert human domain specific knowledge can be the largest driver in the performance improvement, and adversarial learning can be used to further capture nuanced features.  ...  Because generating Pareto optimal plans for the patient requires for the network to learn how to map many dose distributions with tradeoffs from a single anatomy, the neural network must learn to differentiate  ... 
arXiv:1908.05874v2 fatcat:4ubnp3yca5g77asb5txirlfepm

A Transferable Adaptive Domain Adversarial Neural Network for Virtual Reality Augmented EMG-Based Gesture Recognition

Ulysse Cote-Allard, Gabriel Gagnon-Turcotte, Angkoon Phinyomark, Kyrre Glette, Erik Scheme, Francois Laviolette, Benoit Gosselin
2021 IEEE transactions on neural systems and rehabilitation engineering  
This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/  ...  Contrarily, online datasets are limited to the exact EMG-based controller used to record them, necessitating the recording of a new dataset for each control method or variant to be tested.  ...  ACKNOWLEDGMENT The authors would like to thank Alexandre Campeau-Lecours for his support, without which this manuscript would not have been possible.  ... 
doi:10.1109/tnsre.2021.3059741 fatcat:5zg7coxrcffnvcrlxd2brawpvu

A Transferable Adaptive Domain Adversarial Neural Network for Virtual Reality Augmented EMG-Based Gesture Recognition

Ulysse Cote-Allard, Gabriel Gagnon-Turcotte, Angkoon Phinyomark, Kyrre Glette, Erik Scheme, Francois Laviolette, Benoit Gosselin
2021 IEEE transactions on neural systems and rehabilitation engineering  
Contrarily, online datasets are limited to the exact EMG-based controller used to record them, necessitating the recording of a new dataset for each control method or variant to be tested.  ...  The ability of the dynamic dataset to serve as a benchmark is leveraged to evaluate the impact of different recalibration techniques for long-term (across-day) gesture recognition, including a novel algorithm  ...  ACKNOWLEDGMENT The authors would like to thank Alexandre Campeau-Lecours for his support, without which this manuscript would not have been possible.  ... 
doi:10.1109/tnsre.2021.3059741 pmid:33591919 fatcat:siyita4hd5fdfprx4qn4jhpgua

A Transferable Adaptive Domain Adversarial Neural Network for Virtual Reality Augmented EMG-Based Gesture Recognition [article]

Ulysse Côté-Allard, Gabriel Gagnon-Turcotte, Angkoon Phinyomark, Kyrre Glette, Erik Scheme, François Laviolette, Benoit Gosselin
2021 arXiv   pre-print
Contrarily, online datasets are limited to the exact EMG-based controller used to record them, necessitating the recording of a new dataset for each control method or variant to be tested.  ...  The ability of the dynamic dataset to serve as a benchmark is leveraged to evaluate the impact of different recalibration techniques for long-term (across-day) gesture recognition, including a novel algorithm  ...  ACKNOWLEDGMENT The authors would like to thank Alexandre Campeau-Lecours for his support, without which this manuscript would not have been possible.  ... 
arXiv:1912.09380v2 fatcat:giv6oocdajawbdsbybbtxnqpia

DomainGAN: Generating Adversarial Examples to Attack Domain Generation Algorithm Classifiers [article]

Isaac Corley, Jonathan Lwowski, Justin Hoffman
2020 arXiv   pre-print
In this paper, three variants of Generative Adversarial Networks (GANs) are optimized to generate domains which have similar characteristics of benign domains, resulting in domains which greatly evade  ...  Domain Generation Algorithms (DGAs) are frequently used to generate numerous domains for use by botnets.  ...  When generating domains for use offensively, it can become costly to assess whether a domain is in fact usable.  ... 
arXiv:1911.06285v3 fatcat:57ntsi4gffec3irfusexziwuqu

Data Augmentation Generative Adversarial Networks [article]

Antreas Antoniou, Amos Storkey, Harrison Edwards
2018 arXiv   pre-print
The model, based on image conditional Generative Adversarial Networks, takes data from a source domain and learns to take any data item and generalise it to generate other within-class data items.  ...  Effective training of neural networks requires much data. In the low-data regime, parameters are underdetermined, and learnt networks generalise poorly.  ...  We show that such a Data Augmentation Generative Adversarial Network (DAGAN) enables effective neural network training even in low-data target domains.  ... 
arXiv:1711.04340v3 fatcat:woz7kgprrzbyblerqhnpkb2koi

Generate To Adapt: Aligning Domains using Generative Adversarial Networks [article]

Swami Sankaranarayanan, Yogesh Balaji, Carlos D. Castillo, Rama Chellappa
2018 arXiv   pre-print
This is in contrast to methods which use the adversarial framework for realistic data generation and retraining deep models with such data.  ...  We accomplish this by inducing a symbiotic relationship between the learned embedding and a generative adversarial network.  ...  Our objective is to learn an embedding map F : X → R d and a prediction function C : R d → L. In this work, both F and C are modeled as deep neural networks.  ... 
arXiv:1704.01705v4 fatcat:qmaiwj277zahhbenkyyrs2lwoi
« Previous Showing results 1 — 15 out of 32,885 results