A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Filters
Unlearning Protected User Attributes in Recommendations with Adversarial Training
[article]
2022
pre-print
Preserved Fulltext
the disclosure of users' protected attributes. ...
Specifically, we incorporate adversarial training into the state-of-the-art MultVAE architecture, resulting in a novel model, Adversarial Variational Auto-Encoder with Multinomial Likelihood (Adv-MultVAE ...
Unlearning Protected User Attributes in Recommendations with Adversarial Training. ...
doi:10.1145/3477495.3531820
arXiv:2206.04500v1
fatcat:xrb2gpu2lnhw3o3ei6s5bmbsde
Survey: Leakage and Privacy at Inference Time
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
We first discuss what leakage is in the context of different data, tasks, and model architectures. ...
We conclude with outstanding challenges and open questions, outlining some promising directions for future research. ...
At Model Level: Machine Unlearning / Forgetting The General Data Protection Regulation (GDPR), [15] , enforced by the European Union in May 2018, is aimed at protecting user privacy. ...
arXiv:2107.01614v1
fatcat:76a724yzkjfvjisrokssl6assa
Learn to Forget: Machine Unlearning via Neuron Masking
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Nowadays, machine learning models, especially neural networks, become prevalent in many real-world applications.These models are trained based on a one-way trip from user data: as long as users contribute ...
To this end, machine unlearning becomes a popular research topic, which allows users to eliminate memorization of their private data from a trained machine learning model.In this paper, we propose the ...
Recently released data protection regulations, e.g., the California Consumer Privacy Act [6] and the General Data Protection Regulation in the European Union [7] , clearly state that users should have ...
arXiv:2003.10933v3
fatcat:spjwz2chi5ecld7wdskt2adtbu
Efficient Attribute Unlearning: Towards Selective Removal of Input Attributes from Feature Representations
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
The particular attributes will be progressively eliminated along with the training procedure towards convergence, while the rest of attributes related to the main task are preserved for achieving competitive ...
In this paradigm, certain attributes will be accurately captured and detached from the learned feature representations at the stage of training, according to their mutual information. ...
Recently, with the widespread application of Generative Adversarial Network [11] , several works use GAN to augment biased real-world datasets with multiple target labels and protected attributes [34 ...
arXiv:2202.13295v2
fatcat:3z5zoubzwfexxatmpepk27327i
Privacy Enhanced Multimodal Neural Representations for Emotion Recognition
2020
PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
We use an adversarial learning paradigm to unlearn the private information present in a representation and investigate the effect of varying the strength of the adversarial component on the primary task ...
by the user. ...
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF, IBM, TRI, or any other Toyota entity. ...
doi:10.1609/aaai.v34i05.6307
fatcat:r6iazmlvjzdavopnx6nlhoc6uq
Human-Imitating Metrics for Training and Evaluating Privacy Preserving Emotion Recognition Models Using Sociolinguistic Knowledge
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2021 pre-print arXiv:2104.08792v1 fatcat:qp2et6kcuffmjpigusndgha2aa
But, in applications relying on machine learning backends, privacy is challenging because models often capture more than what the model was initially trained for, resulting in the potential leakage of ...
We use the degree with which differences in interpretation of general vs privacy preserving models correlate with sociolinguistic biases to inform metric design. ...
Privacy by Adversarial Training We use an adversarial paradigm to train models to preserve privacy of the generated embeddings with respect to the demographic variable of gender. ...
arXiv:2104.08792v2
fatcat:irogim2gbbhkzcvhqt643jgv4e
Zero-Shot Machine Unlearning
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We therefore ask the question: is it possible to achieve unlearning with zero training samples? ...
Thus, in many cases, no data related to the training process or training samples may be accessible even for the unlearning purpose. ...
Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of National Research Foundation, Singapore. ...
arXiv:2201.05629v2
fatcat:nhpz3562qrgivloisvbwy4hcsa
How to keep text private? A systematic review of deep learning methods for privacy-preserving natural language processing
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Data protection laws, such as the European Union's General Data Protection Regulation (GDPR), thereby enforce the need for privacy. ...
Third, throughout the review, we describe privacy issues in the NLP pipeline in a holistic view. ...
Statements and declarations All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter ...
arXiv:2205.10095v1
fatcat:rksy7oxxlbde5bol3ay44yycru
Machine Learning Security: Threats, Countermeasures, and Evaluations
2020
IEEE Access
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Then, the machine learning security-related issues are classified into five categories: training set poisoning; backdoors in the training set; adversarial example attacks; model theft; recovery of sensitive ...
It has demonstrated significant success in dealing with various complex problems, and shows capabilities close to humans or even beyond humans. ...
They generate fake users with crafted rating scores based on an optimization problem, and inject them to the recommender system. ...
doi:10.1109/access.2020.2987435
fatcat:ksinvcvcdvavxkzyn7fmsa27ji
Backdoor Attacks and Countermeasures on Deep Learning: A Comprehensive Review
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
In some cases, an attacker can intelligently bypass existing defenses with an adaptive attack. ...
We have also reviewed the flip side of backdoor attacks, which are explored for i) protecting intellectual property of deep learning models, ii) acting as a honeypot to catch adversarial example attacks ...
In this case, the server needs to unlearn the data from this user. One intuitive way is to retrain the model from scratch after removing the user data as requested. ...
arXiv:2007.10760v3
fatcat:6i4za6345jedbe5ek57l2tgld4
SoK: Managing Longitudinal Privacy of Publicly Shared Personal Online Data
2021
Proceedings on Privacy Enhancing Technologies
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
While limitations of data availability in proposed approaches and real systems are mostly time-based, users' desired models are rather complex, taking into account content, audience, and the context in ...
In this work, we systematize research on longitudinal privacy management of publicly shared personal online data from these two perspectives: user studies capturing users' interactions related to the availability ...
When the data is used to aggregate statistics or to train machine-learning models, e. g., for image classification or recommender systems, the information that data carries will implicitly remain in the ...
doi:10.2478/popets-2021-0013
fatcat:5ehzkvuhbvhend3mf5xd2tl434
Bias in Multimodal AI: Testbed for Fair Automatic Recruitment
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
We train automatic recruitment algorithms using a set of multimodal synthetic profiles consciously scored with gender and racial biases. ...
With the aim of studying how current multimodal algorithms based on heterogeneous sources of information are affected by sensitive elements and inner biases in the data, we propose a fictitious automated ...
datasets that approximate a given original one, but more fair with respect to certain protected attributes. ...
arXiv:2004.07173v1
fatcat:x6zbp3jd7bdwvptwtkiorisrya
Bias in Multimodal AI: Testbed for Fair Automatic Recruitment
2020
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
We train automatic recruitment algorithms using a set of multimodal synthetic profiles consciously scored with gender and racial biases. ...
With the aim of studying how current multimodal algorithms based on heterogeneous sources of information are affected by sensitive elements and inner biases in the data, we propose a fictitious automated ...
datasets that approximate a given original one, but more fair with respect to certain protected attributes. ...
doi:10.1109/cvprw50498.2020.00022
dblp:conf/cvpr/PenaSMF20
fatcat:7fbnokkqfngbvn5yneac4b2amm
A Prompt Array Keeps the Bias Away: Debiasing Vision-Language Models with Adversarial Learning
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We address both of these challenges in this paper: First, we evaluate different bias measures and propose the use of retrieval metrics to image-text representations via a bias measuring framework. ...
Second, we investigate debiasing methods and show that optimizing for adversarial loss via learnable token embeddings minimizes various bias measures without substantially degrading feature representations ...
Model-based debiasing methods are more similar to our work, these include optimizing confusion [4] , domain adversarial training [18] , or training a network to unlearn bias information [27] . ...
arXiv:2203.11933v2
fatcat:d5ahgqbrmbhlndyr6ln5tjvbwa
Adaptive Machine Unlearning
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Combined with ideas from prior work which give guarantees for non-adaptive deletion sequences, this leads to extremely flexible algorithms able to handle arbitrary model classes and training methodologies ...
Data deletion algorithms aim to remove the influence of deleted data points from trained models at a cheaper computational cost than fully retraining those models. ...
against adaptive adversarial streams to streaming algorithms with guarantees against oblivious adversaries. ...
arXiv:2106.04378v1
fatcat:lgiq7qk2kvgonln3ktpafnlrx4
« Previous
Showing results 1 — 15 out of 302 results