2,378 Hits in 3.6 sec

Subgroup Generalization and Fairness of Graph Neural Networks [article]

Jiaqi Ma, Junwei Deng, Qiaozhu Mei
2021 arXiv   pre-print
Despite enormous successful applications of graph neural networks (GNNs), theoretical understanding of their generalization ability, especially for node-level tasks where data are not independent and identically-distributed  ...  The theoretical investigation of the generalization performance is beneficial for understanding fundamental issues (such as fairness) of GNN models and designing better learning methods.  ...  [37] Saurabh Verma and Zhi-Li Zhang. Stability and generalization of graph convolutional neural networks.  ... 
arXiv:2106.15535v2 fatcat:zxlgget6vzekxkghwkqjlbc4aq

Say No to the Discrimination: Learning Fair Graph Neural Networks with Limited Sensitive Attribute Information [article]

Enyan Dai, Suhang Wang
2021 arXiv   pre-print
Graph neural networks (GNNs) have shown great power in modeling graph structured data.  ...  Therefore, we study the novel and important problem of learning fair GNNs with limited sensitive attribute information.  ...  The findings and conclusions in this paper do not necessarily reflect the view of the funding agency.  ... 
arXiv:2009.01454v5 fatcat:wb2bqq4khfcxfim4k7h2fxkpou

The KL-Divergence between a Graph Model and its Fair I-Projection as a Fairness Regularizer [article]

Maarten Buyl, Tijl De Bie
2021 arXiv   pre-print
Learning and reasoning over graphs is increasingly done by means of probabilistic models, e.g. exponential random graph models, graph embedding models, and graph neural networks.  ...  Given this, we propose a fairness regularizer defined as the KL-divergence between the graph model and its I-projection onto the set of fair models.  ...  of Ghent University (PhD scholarship BOF20/DOC/144), and the FWO (project no.  ... 
arXiv:2103.01846v2 fatcat:k2pexhdubbdjjdfepypyvj5buy

Childhood maltreatment is associated with a sex-dependent functional reorganization of a brain inhibitory control network

Amanda Elton, Shanti P. Tripathi, Tanja Mletzko, Jonathan Young, Josh M. Cisler, G. Andrew James, Clinton D. Kilts
2013 Human Brain Mapping  
Graph theoretical analyses and structural equation modeling investigated the impact of childhood maltreatment on the functional organization of this neural processing network.  ...  Graph theory outcomes revealed sex differences in the relationship between network functional connectivity and inhibitory control which were dependent on the severity of childhood maltreatment exposure  ...  ACKNOWLEDGMENTS The authors thank Tim Ely for fMRI stop-signal task design and Kristina Davidson for her assistance with data entry.  ... 
doi:10.1002/hbm.22280 pmid:23616424 pmcid:PMC3779516 fatcat:2zvj662bkbeqberoerewtqww6m

Model-based network discovery of developmental and performance-related differences during risky decision-making

Ethan M. McCormick, Kathleen M. Gates, Eva H. Telzer
2019 NeuroImage  
Theories of adolescent neurodevelopment have largely focused on group-level descriptions of neural changes that help explain increases in risk behavior that are stereotypical of the teen years.  ...  Here we apply GIMME, a model-based approach which uses both group and individual-level information to construct functional connectivity maps, to investigate risky behavior and neural changes across development  ...  We greatly appreciate the assistance of the Biomedical Imaging Center at the University of Illinois.  ... 
doi:10.1016/j.neuroimage.2018.12.042 pmid:30579902 pmcid:PMC6401275 fatcat:ztg2yqquizfb5blnjoda5f4ihy

Developmental Maturation of the Precuneus as a Functional Core of the Default Mode Network

Rosa Li, Amanda V. Utevsky, Scott A. Huettel, Barbara R. Braams, Sabine Peters, Eveline A. Crone, Anna C. K. van Duijvenvoorde
2019 Journal of Cognitive Neuroscience  
Efforts to map the functional architecture of the developing human brain have shown that connectivity between and within functional neural networks changes from childhood to adulthood.  ...  By applying independent component analysis to both task and rest data, we identified three canonical networks of interest-the rest-based default mode network and the task-based left and right frontoparietal  ...  We generated bidirectional contrasts comparing task and rest states in each of the three networks of interest for each task + rest subgroup at each longitudinal time point.  ... 
doi:10.1162/jocn_a_01426 pmid:31112473 fatcat:g4e5p7xl75ctxflrepm3a7jwxa

Fairness without the sensitive attribute via Causal Variational Autoencoder [article]

Vincent Grari, Sylvain Lamprier, Marcin Detyniecki
2021 arXiv   pre-print
We observe that the generated proxy's latent space recovers sensitive information and that our approach achieves a higher accuracy while obtaining the same level of fairness on two real datasets, as measured  ...  We notice a lack of approaches for mitigating bias in such difficult settings, in particular for achieving classical fairness objectives such as Demographic Parity and Equalized Odds.  ...  The neural network f with parameters w f takes as input the variable U and the neural network g with parameters w g takes as input V .  ... 
arXiv:2109.04999v1 fatcat:4hjxnhhvtbblbdyr2wt5buthjq

Developmental maturation of the precuneus as a functional core of the default-mode network [article]

Rosa Li, Amanda Utevsky, Scott Huettel, Barbara Braams, Sabine Peters, Eveline Crone, Anna C.K. van Duijvenvoorde
2018 bioRxiv   pre-print
Efforts to map the functional architecture of the developing human brain have shown that connectivity between and within functional neural networks changes from childhood to adulthood.  ...  By applying independent component analysis (ICA) to both task and rest data, we identified three canonical networks of interest -- the rest-based default mode network (DMN) and the task-based left and  ...  We generated bidirectional contrasts comparing task and rest states in each of the three networks of interest for each task+rest subgroup at each longitudinal timepoint.  ... 
doi:10.1101/419028 fatcat:23gns4owtvdj7g74e2ok5ng4c4

Differences in cognitive aging: typology based on a community structure detection approach

Emi Saliasi, Linda Geerligs, Jelle R. Dalenberg, Monicque M. Lorist, Natasha M. Maurits
2015 Frontiers in Aging Neuroscience  
An important novelty of this study is the use of graph-based community structure detection analysis to map performance in a mixed population of 79 young and 76 older adults, without separating the age  ...  In addition, we identified a subgroup of young and older adults who performed at a similar cognitive level of overall good cognitive performance with slightly decreased processing speed.  ...  Park and Reuter-Lorenz (2009) also elaborate on the efficient recruitment of additional neural networks, which they call scaffolding networks or scaffolds (Park and Reuter-Lorenz, 2009 ).  ... 
doi:10.3389/fnagi.2015.00035 pmid:25852549 pmcid:PMC4365722 fatcat:dmu7iyspavaxnktdyif4nvq5ya

Modification-Fair Cluster Editing [article]

Vincent Froese, Leon Kellerhals, Rolf Niedermeier
2021 arXiv   pre-print
We complement these and further theoretical results with an empirical analysis of our model on real-world social networks where we find that the price of modification-fairness is surprisingly low, that  ...  However, in the more general editing form, the modification-fair variant remains fixed-parameter tractable with respect to the number of edge edits.  ...  For small graphs, the modification fairness is rather low, with ∆norm ≥ 0.1 for 35% of the graphs in Sets 1 and 2.  ... 
arXiv:2112.03183v1 fatcat:zdq24phxhrexxppuqzsej5prya

Subgroup Fairness in Graph-based Spam Detection [article]

Jiaxin Liu, Yuefei Lyu, Xi Zhang, Sihong Xie
2022 arXiv   pre-print
This paper addresses the challenges of defining, discovering, and utilizing subgroup memberships for fair spam detection.  ...  The complex dependencies over the review graph introduce difficulties in teasing out subgroups of reviewers that are hidden within larger groups and are treated unfairly.  ...  For those non-Euclidean data, Graph Neural Networks (GNNs) are powerful architectures for graph representation and learning.  ... 
arXiv:2204.11164v1 fatcat:osq6ii67tzbgzj3kdpmk5beeiq

Partition-Based Active Learning for Graph Neural Networks [article]

Jiaqi Ma, Ziqiao Ma, Joyce Chai, Qiaozhu Mei
2022 arXiv   pre-print
We study the problem of semi-supervised learning with Graph Neural Networks (GNNs) in an active learning setup. We propose GraphPart, a novel partition-based active learning approach for GNNs.  ...  The proposed method is motivated by a novel analysis of the classification error under realistic smoothness assumptions over the graph and the node features.  ...  [Ma et al., 2021] Jiaqi Ma, Junwei Deng, and Qiaozhu Mei. Subgroup generalization and fairness of graph neural net-works. In Advances in Neural Information Processing Sys-tems (NeurIPS) 34.  ... 
arXiv:2201.09391v1 fatcat:k45ceaufu5hvfcvzzflv46dpgq

Distraction is All You Need for Fairness [article]

Mehdi Yazdani-Jahromi and AmirArsalan Rajabi and Aida Tayebi and Ozlem Ozmen Garibay
2022 arXiv   pre-print
We compare our model with six state-of-the-art methodologies proposed in fairness literature, and show that the model is superior to those methods in terms of minimizing bias while maintaining accuracy  ...  In this paper, we propose a novel classification algorithm that improves fairness, while maintaining accuracy of the predictions.  ...  Acknowledgment We thank Negin Khoeiniha for proof checking the Proposition 2.1 and providing insights on the proof.  ... 
arXiv:2203.07593v1 fatcat:3eyyaje6czabzo5p4fx6g3oji4

Bursting the Filter Bubble: Fairness-Aware Network Link Prediction

Farzan Masrour, Tyler Wilson, Heng Yan, Pang-Ning Tan, Abdol Esfahanian
In this study, we examine the filter bubble problem from the perspective of algorithm fairness and introduce a dyadic-level fairness criterion based on network modularity measure.  ...  Experimental results conducted on several real-world datasets showed the effectiveness of the proposed methods compared to other baseline approaches, which include conventional link prediction and fairness-aware  ...  Given a graph N = (V, E, X) we generate a training set with equal number of negative and positive examples < N , E + , E − >.  ... 
doi:10.1609/aaai.v34i01.5429 fatcat:zm5apbexozgnhcso53yu3kckhm

A Survey on Fairness for Machine Learning on Graphs [article]

Manvi Choudhary and Charlotte Laclau and Christine Largeron
2022 arXiv   pre-print
It aims to present a comprehensive review of state-of-the-art techniques in fairness on graph mining and identify the open challenges and future trends.  ...  In that context, algorithmic contributions for graph mining are not spared by the problem of fairness and present some specific challenges related to the intrinsic nature of graphs: (1) graph data is non-IID  ...  Graph Neural Networks based models Graph Neural Networks (GNN) models have recently established themselves as state-of-the-art for the graph-related downstream tasks.  ... 
arXiv:2205.05396v1 fatcat:t7nis5olbretdf3nwarsho5x5i
« Previous Showing results 1 — 15 out of 2,378 results