60,299 Hits in 6.6 sec

Generalized Adversarially Learned Inference [article]

Yatin Dandi, Homanga Bharadhwaj, Abhishek Kumar, Piyush Rai
2020 arXiv   pre-print
Recent approaches, such as ALI and BiGAN frameworks, develop methods of inference of latent variables in GANs by adversarially training an image generator along with an encoder to match two joint distributions  ...  We generalize these approaches to incorporate multiple layers of feedback on reconstructions, self-supervision, and other forms of supervision based on prior or learned knowledge about the desired solutions  ...  In the rest of the paper, we refer to our proposed framework as Generalized Adversarially Learned Inference (GALI).  ... 
arXiv:2006.08089v3 fatcat:2ji66xyarzd4peuzetlesk2tei

Generative Adversarial Reward Learning for Generalized Behavior Tendency Inference [article]

Xiaocong Chen, Lina Yao, Xianzhi Wang, Aixin Sun, Wenjie Zhang, Quan Z. Sheng
2021 arXiv   pre-print
We propose a generative inverse reinforcement learning for user behavioral preference modelling, to address the above issues.  ...  Besides, they generally use task-specific reward functions that sacrifice generalization ability.  ...  Policy generation is the problem of matching two occupancy measures and can be solved by training a Generative Adversarial Network (GAN) [28] .  ... 
arXiv:2105.00822v2 fatcat:itlyo4txfjbodp3wyb7oqehg3a

Adversarial Learned Molecular Graph Inference and Generation [article]

Sebastian Pölsterl, Christian Wachinger
2020 arXiv   pre-print
In this work, we propose ALMGIG, a likelihood-free adversarial learning framework for inference and de novo molecule generation that avoids explicitly computing a reconstruction loss.  ...  Our approach extends generative adversarial networks by including an adversarial cycle-consistency loss to implicitly enforce the reconstruction property.  ...  Conclusion We formulated generation and inference of molecular graphs as a likelihood-free adversarial learning task.  ... 
arXiv:1905.10310v2 fatcat:lpephwlkl5cqnklxck44wxygya

Learning and inference on generative adversarial quantum circuits

Jinfeng Zeng, Yufeng Wu, Jin-Guo Liu, Lei Wang, Jiangping Hu
2019 Physical Review A  
We numerically simulate the learning and inference of generative adversarial quantum circuit using the prototypical Bars-and-Stripes dataset.  ...  Generative adversarial quantum circuits is a fresh approach to machine learning which may enjoy the practically useful quantum advantage on near-term quantum devices.  ...  ACKNOWLEDGMENT Learning and inference of the generative adversarial quantum circuits are implemented using Yao.jl  ... 
doi:10.1103/physreva.99.052306 fatcat:7ctrtz4xcjf5npstoxreye7wua

Evaluating and Mitigating Bias in Image Classifiers: A Causal Perspective Using Counterfactuals [article]

Saloni Dash, Vineeth N Balasubramanian, Amit Sharma
2022 arXiv   pre-print
We present a method for generating counterfactuals by incorporating a structural causal model (SCM) in an improved variant of Adversarially Learned Inference (ALI), that generates counterfactuals in accordance  ...  Based on the generated counterfactuals, we show how to explain a pre-trained machine learning classifier, evaluate its bias, and mitigate the bias using a counterfactual regularizer.  ...  Implementing the Encoder and Generator Adversarially Learned Inference.  ... 
arXiv:2009.08270v4 fatcat:whtgbxkrprgn7oyx3gxayqnhae

Adversarial Message Passing For Graphical Models [article]

Theofanis Karaletsos
2016 arXiv   pre-print
A popular tool for learning implicit models are generative adversarial networks (GANs) which learn parameters of generators by fooling discriminators.  ...  This allows us to compose models and yields a unified inference and learning framework for adversarial learning.  ...  We furthermore acknowledge Anh Nguyen and Jason Yosinski for empirical demonstrations of the benefits of adversarial learning.  ... 
arXiv:1612.05048v1 fatcat:fy4oxhajengvhbofxmophj3e6e

Generative Adversarial Networks for Black-Box API Attacks with Limited Training Data [article]

Yi Shi, Yalin E. Sagduyu, Kemal Davaslioglu, Jason H. Li
2019 arXiv   pre-print
In return, a generative adversarial network (GAN) based on deep learning is built to generate synthetic training data from a limited number of real training data samples, thereby extending the training  ...  The adversary launches an exploratory (inference) attack by querying the API of an online machine learning system (in particular, a classifier) with input data samples, collecting returned labels to build  ...  Index Terms-Adversarial machine learning, exploratory attack, causative attack, evasion attack, deep learning, generative adversarial network. I.  ... 
arXiv:1901.09113v1 fatcat:xubnsti6obaitjzd7lzz2b2yvm

Towards Demystifying Membership Inference Attacks [article]

Stacey Truex, Ling Liu, Mehmet Emre Gursoy, Lei Yu, Wenqi Wei
2019 arXiv   pre-print
Membership inference attacks seek to infer membership of individual training instances of a model to which an adversary has black-box access through a machine learning-as-a-service API.  ...  to membership inference risks when the adversary is a participant.  ...  Membership Inference v.s. Adversarial Examples. Adversarial learning to-date has been focused on attacking deployed deep learning models.  ... 
arXiv:1807.09173v2 fatcat:5vmtqv5glndphomymzpm2k2rpu

A GAN-based Approach for Mitigating Inference Attacks in Smart Home Environment [article]

Olakunle Ibitoye, Ashraf Matrawy, M. Omair Shafiq
2020 arXiv   pre-print
We propose a Generative Adversarial Network (GAN) based approach for privacy preservation in smart homes which generates random noise to distort the unwanted machine learning-based inference.  ...  In this study, we explore the problem of adversaries spying on smart home users to infer sensitive information with the aid of machine learning techniques.  ...  Our solution is based on a generative deep learning model known as the Generative Adversarial Networks (GAN).  ... 
arXiv:2011.06725v1 fatcat:5m54yleul5g6pmpxb4f37kydvi

Is Homomorphic Encryption-Based Deep Learning Secure Enough?

Jinmyeong Shin, Seok-Hwan Choi, Yoon-Ho Choi
2021 Sensors  
From the experimental evaluation results, we show that the adversarial example and reconstruction attacks are a practical threat to homomorphic encryption-based deep learning models.  ...  ; (2) a reconstruction attack using the paired input and output data; and (3) a membership inference attack by malicious insider.  ...  The adversarial attack is a process to generate human-imperceptible perturbations and there are three well-known adversarial attack methods.  ... 
doi:10.3390/s21237806 pmid:34883809 fatcat:j5rqqan2irhn7ndoomjc4hb7yy

Privacy Risks of Securing Machine Learning Models against Adversarial Examples

Liwei Song, Reza Shokri, Prateek Mittal
2019 Proceedings of the 2019 ACM SIGSAC Conference on Computer and Communications Security - CCS '19  
We also propose two new inference methods that exploit structural properties of robust models on adversarially perturbed data.  ...  The arms race between attacks and defenses for machine learning models has come to a forefront in recent years, in both the security community and the privacy community.  ...  In particular, we seek to understand the privacy risks of securing machine learning models by evaluating membership inference attacks against adversarially robust deep learning models, which aim to mitigate  ... 
doi:10.1145/3319535.3354211 dblp:conf/ccs/SongSM19 fatcat:32ckh3h7gnfw3hphzyhyy3cgty

Infer-AVAE: An Attribute Inference Model Based on Adversarial Variational Autoencoder [article]

Yadong Zhou, Zhihao Ding, Xiaoming Liu, Chao Shen, Lingling Tong, Xiaohong Guan
2021 arXiv   pre-print
In this paper, we propose an attribute Inference model based on Adversarial VAE (Infer-AVAE) to cope with these issues.  ...  Specifically, to overcome over-smoothing, Infer-AVAE unifies MLP and GNNs in encoder to learn positive and negative latent representations respectively.  ...  Inspired by Generative Adversarial Network (GAN) [15] , our model equips VAE with adversarial network to relieve over-smoothing.  ... 
arXiv:2012.15005v2 fatcat:rfawvt3vsvbj5lrrnx2wtmidaa

Quantifying and Mitigating Privacy Risks of Contrastive Learning [article]

Xinlei He, Yang Zhang
2021 arXiv   pre-print
To remedy this situation, we propose the first privacy-preserving contrastive learning mechanism, Talos, relying on adversarial training.  ...  In this way, a contrastive model learns to generate informative representations for data samples, which are then used to perform downstream ML tasks.  ...  Here, an adversary aims to infer a specific sensitive attribute of a data sample from its representation generated by a target model [36, 56] .  ... 
arXiv:2102.04140v2 fatcat:nbl33eaitnd75ontpjcw65jkia

Threats to Federated Learning: A Survey [article]

Lingjuan Lyu, Han Yu, Qiang Yang
2020 arXiv   pre-print
Federated learning (FL) has recently emerged as a promising solution under this new reality.  ...  Existing FL protocol design has been shown to exhibit vulnerabilities which can be exploited by adversaries both within and without the system to compromise data privacy.  ...  Inference attacks generally assume that the adversaries possess sophisticated technical capabilities and large computational resources.  ... 
arXiv:2003.02133v1 fatcat:htv4tztwlbdihdkat5bzlcm46y

Privacy Risk in Machine Learning: Analyzing the Connection to Overfitting [article]

Samuel Yeom, Irene Giacomelli, Matt Fredrikson, Somesh Jha
2018 arXiv   pre-print
inference or attribute inference attacks.  ...  Machine learning algorithms, when applied to sensitive data, pose a distinct threat to privacy.  ...  Attribute inference and reduction We now present the empirical attribute advantage of the general adversary (Adversary 4).  ... 
arXiv:1709.01604v5 fatcat:facq7utbrrdghaqfeveesswniy
« Previous Showing results 1 — 15 out of 60,299 results