63 research outputs found
Explaining Graph Neural Networks
No full textGraph Neural Networks are an up-and-coming class of neural networks that operate on graphs and can therefore deal with connected, highly complex data. As explaining neural networks becomes more and more important, we investigate different ways to explain graph neural networks and contrast gradient-based explanations with the interpretability by design approach KEdge.
We extend KEdge, to work with probability distributions different from HardKuma. Our goal
is to test the performance of each method to judge which one works best under given circum-
stances. For this, we extend the notion of fidelity from hard attribution weights to soft attribution
weights and use the resulting metric to evaluate the explanations generated by KEdge, as well
as by the gradient-based techniques.
We also compare the predictive performance of models that use KEdge with different distributions. Our experiments are run on the Cora, SightSeer, Pubmed, and MUTAG datasets. We find that KEdge outperforms the gradient based attribution techniques on graph classification problems and that it should be used with the HardNormal, HardKuma, or HardLaplace distributions, depending on if the top priority is model performance or attribution quality. To compare different metrics of judging attributions in the text domain, we visualize attribution weights generated by different models and find, that metrics which
compare model attributions to human explanations lead to bad attribution weights
Explaining and Applying Graph Neural Networks on Text
No full textText classification is an essential task in natural language processing. While graph neural networks (GNNs) have successfully been applied to this problem both through graph classification and node classification approaches, their typical applications suffer from several issues. In the graph classification case, common graph construction techniques tend to leave out syntactic information. In the node classification case, most widespread datasets and applications tend to suffer from encoding relatively little information in the chosen node features. Finally, there are great benefits to be gained from combining the two GNN approaches. To tackle these concerns, we propose DepNet, a two-stage framework for text classification using GNN models. In the first stage we replace current graph construction methods by utilizing syntactic dependency parsing in order to include as much syntactic information in the GNN input as possible. In the second stage we combine both graph classification and node classification methods by utilizing the former to produce node embeddings for the latter, maximizing the potential of GNNs for text classification. We find that this technique significantly improves the performance of both graph classification and node classification approaches to text classification
The smashHitCore ontology for GDPR-compliant sensor data sharing in smart cities
No full textThe adoption of the General Data Protection Regulation (GDPR) has resulted in a significant shift in how the data of European Union citizens is handled. A variety of data sharing challenges in scenarios such as smart cities have arisen, especially when attempting to semantically represent GDPR legal bases, such as consent, contracts and the data types and specific sources related to them. Most of the existing ontologies that model GDPR focus mainly on consent. In order to represent other GDPR bases, such as contracts, multiple ontologies need to be simultaneously reused and combined, which can result in inconsistent and conflicting knowledge representation. To address this challenge, we present the smashHitCore ontology. smashHitCore provides a unified and coherent model for both consent and contracts, as well as the sensor data and data processing associated with them. The ontology was developed in response to real-world sensor data sharing use cases in the insurance and smart city domains. The ontology has been successfully utilised to enable GDPR-complaint data sharing in a connected car for insurance use cases and in a city feedback system as part of a smart city use case
Analyzing and Predicting Material Flow Networks Using Stochastic Block Models and Statistical Graph Embeddings
No full textManufacturing and logistics systems consist of many complexly interacting elements. Starting from social science, the field of complex networks has developed concepts and methods to analyze and predict networks, such as friendship networks or protein interactions. However, although these examples have equivalents in the form of company networks and interactions within manufacturing processes, more sophisticated methods have not yet been transferred to manufacturing and logistics. We propose to apply methods from clustering and graph embedding on representations of machine interactions to analyze the structural stability of manufacturing systems and to predict structural changes of such systems
Zorro: Valid, sparse, and stable explanations in graph neural networks
No full textWith the ever-increasing popularity and applications of graph neural networks, several proposals have been made to explain and understand the decisions of a graph neural network. Explanations for graph neural networks differ in principle from other input settings. It is important to attribute the decision to input features and other related instances connected by the graph structure. We find that the previous explanation generation approaches that maximize the mutual information between the label distribution produced by the model and the explanation to be restrictive. Specifically, existing approaches do not enforce explanations to be valid, sparse, or robust to input perturbations. In this paper, we lay down some of the fundamental principles that an explanation method for graph neural networks should follow and introduce a metric RDT-Fidelity as a measure of the explanation's effectiveness. We propose a novel approach Zorro based on the principles from rate-distortion theory that uses a simple combinatorial procedure to optimize for RDT-Fidelity. Extensive experiments on real and synthetic datasets reveal that Zorro produces sparser, stable, and more faithful explanations than existing graph neural network explanation approaches.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Multimedia ComputingWeb Information System
Stochastic block models: A comparison of variants and inference methods.
No full textFinding communities in complex networks is a challenging task and one promising approach is the Stochastic Block Model (SBM). But the influences from various fields led to a diversity of variants and inference methods. Therefore, a comparison of the existing techniques and an independent analysis of their capabilities and weaknesses is needed. As a first step, we review the development of different SBM variants such as the degree-corrected SBM of Karrer and Newman or Peixoto's hierarchical SBM. Beside stating all these variants in a uniform notation, we show the reasons for their development. Knowing the variants, we discuss a variety of approaches to infer the optimal partition like the Metropolis-Hastings algorithm. We perform our analysis based on our extension of the Girvan-Newman test and the Lancichinetti-Fortunato-Radicchi benchmark as well as a selection of some real world networks. Using these results, we give some guidance to the challenging task of selecting an inference method and SBM variant. In addition, we give a simple heuristic to determine the number of steps for the Metropolis-Hastings algorithms that lack a usual stop criterion. With our comparison, we hope to guide researches in the field of SBM and highlight the problem of existing techniques to focus future research. Finally, by making our code freely available, we want to promote a faster development, integration and exchange of new ideas
A Tool for an Analysis of the Dynamic Behavior of Logistic Systems with the Instruments of Complex Networks
No full textComplex networks of material flow in manufacturing and logistics: Modeling, analysis, and prediction using stochastic block models
No full textPrivate Graph Extraction via Feature Explanations
No full textPrivacy and interpretability are two important ingredients for achieving trustworthy machine learning. We study the interplay of these two aspects in graph machine learning through graph reconstruction attacks. The goal of the adversary here is to reconstruct the graph structure of the training data given access to model explanations. Based on the different kinds of auxiliary information available to the adversary, we propose several graph reconstruction attacks. We show that additional knowledge of post-hoc feature explanations substantially increases the success rate of these attacks. Further, we investigate in detail the differences between attack performance with respect to three different classes of explanation methods for graph neural networks: gradient-based, perturbationbased, and surrogate model-based methods. While gradient-based explanations reveal the most in terms of the graph structure, we find that these explanations do not always score high in utility. For the other two classes of explanations, privacy leakage increases with an increase in explanation utility. Finally, we propose a defense based on a randomized response mechanism for releasing the explanations, which substantially reduces the attack success rate. Our code is available at https://github.com/iyempissy/graphstealing- attacks-with-explanation.Multimedia Computin
Animating Still Images: Folding Texture Design and Synthesis
No full textThe phenomenon of one element moving and progressively overlaying another is common in nature, such as waves swashing and backwashing, or eyelids moving over eyeballs while blinking. Folding Texture, which was proposed by Thorben, can simulate this texture “folding” visual effect in real-time without changing geometry.However, to date, no tool has been developed to assist in the design and synthesis of folding textures. Applications of the technique so far are achieved through manual creation of the folding texture, which is a tedious process.This thesis explores the problem of folding-texture design and synthesis. A novel approach is proposed for animating still images based on the folding texture technique. The approach uses a semi-automatic, user-assisted method that combines texture editing, motion profile specification, and folding texture synthesis into one seamless process, reducing the need for extensive manual work. It enables novice users to utilize the technique with a fair level of prior knowledge of folding texture.Computer Scienc
- …
