An official website of the United States government
This work considers the task of representation learning on the attributed relational graph (ARG). Both the nodes and edges in an ARG are associated with attributes/features allowing ARGs to encode rich structural information widely observed in real applications. Existing graph neural networks offer limited ability to capture complex interactions within local structural contexts, which hinders them from taking advantage of the expression power of ARGs. We propose motif convolution module (MCM), a new motif-based graph representation learning technique to better utilize local structural information. The ability to handle continuous edge and node features is one of MCM’s advantages over existing motif-based models. MCM builds a motif vocabulary in an unsupervised way and deploys a novel motif convolution operation to extract the local structural context of individual nodes, which is then used to learn higher level node representations via multilayer perceptron and/or message passing in graph neural networks. When compared with other graph learning approaches to classifying synthetic graphs, our approach is substantially better at capturing structural context. We also demonstrate the performance and explainability advantages of our approach by applying it to several molecular benchmarks.
more »
« less
Topological Detection of Trojaned Neural Networks
Deep neural networks are known to have security issues. One particular threat is the Trojan attack. It occurs when the attackers stealthily manipulate the model's behavior through Trojaned training samples, which can later be exploited. Guided by basic neuroscientific principles we discover subtle -- yet critical -- structural deviation characterizing Trojaned models. In our analysis we use topological tools. They allow us to model high-order dependencies in the networks, robustly compare different networks, and localize structural abnormalities. One interesting observation is that Trojaned models develop short-cuts from input to output layers. Inspired by these observations, we devise a strategy for robust detection of Trojaned models. Compared to standard baselines it displays better performance on multiple benchmarks.
more »
« less
- Award ID(s):
- 1910873
- PAR ID:
- 10366262
- Date Published:
- Journal Name:
- Advances in neural information processing systems
- ISSN:
- 1049-5258
- Page Range / eLocation ID:
- 17258--17272
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract In many applications, it is of interest to identify anomalous behavior within a dynamic interacting system. Such anomalous interactions are reflected by structural changes in the network representation of the system. We propose and investigate the use of the degree corrected stochastic block model (DCSBM) to model and monitor dynamic networks that undergo a significant structural change. We apply statistical process monitoring techniques to the estimated parameters of the DCSBM to identify significant structural changes in the network. We apply our surveillance strategy to a dynamic US Senate covoting network. We detect significant changes in the political network that reflect both times of cohesion and times of polarization among Republican and Democratic party members. Our analysis demonstrates that the DCSBM monitoring procedure effectively detects local and global structural changes in complex networks, providing useful insights into the modeled system. The DCSBM approach is an example of a general framework that combines parametric random graph models and statistical process monitoring techniques for network surveillance.more » « less
-
One of the central elements of any causal inference is an object called structural causal model (SCM), which represents a collection of mechanisms and exogenous sources of random variation of the system under investigation (Pearl, 2000). An important property of many kinds of neural networks is universal approximability: the ability to approximate any function to arbitrary precision. Given this property, one may be tempted to surmise that a collection of neural nets is capable of learning any SCM by training on data generated by that SCM. In this paper, we show this is not the case by disentangling the notions of expressivity and learnability. Specifically, we show that the causal hierarchy theorem (Thm. 1, Bareinboim et al., 2020), which describes the limits of what can be learned from data, still holds for neural models. For instance, an arbitrarily complex and expressive neural net is unable to predict the effects of interventions given observational data alone. Given this result, we introduce a special type of SCM called a neural causal model (NCM), and formalize a new type of inductive bias to encode structural constraints necessary for performing causal inferences. Building on this new class of models, we focus on solving two canonical tasks found in the literature known as causal identification and estimation. Leveraging the neural toolbox, we develop an algorithm that is both sufficient and necessary to determine whether a causal effect can be learned from data (i.e., causal identifiability); it then estimates the effect whenever identifiability holds (causal estimation). Simulations corroborate the proposed approach.more » « less
-
na (Ed.)Over the last two decades, there has been a growth in the applications of geographically-explicit agent-based models. One thing such models have in common is the creation of synthetic populations to initialize the artificial worlds in which the agents inhabit. One challenge such models face is that it is often difficult to create reusable geographically-explicit synthetic populations with social networks. In this paper, we introduce a Python based method that generates a reusable geographically-explicit synthetic population dataset along with its social networks. In addition, we present a pipeline for using the population datasets for model initialization. With this pipeline, multiple spatial and temporal scales of geographically-explicit agent-based models are presented focusing on Western New York. Such models not only demonstrate the utility of our synthetic population on commuting patterns but also how social networks can impact the simulation of disease spread and vaccination uptake. By doing so, this pipeline could benefit any modeler wishing to reuse synthetic populations with realistic geographic locations and social networks.more » « less
-
Detecting Cracks and Spalling Automatically in Extreme Events by End-to-end Deep Learning FrameworksIn this paper, we develop and implement end-to-end deep learning approaches to automatically detect two important types of structural failures, cracks and spalling, of buildings and bridges in extreme events such as major earthquakes. A total of 2,229 images were annotated, and are used to train and validate three newly developed Mask Regional Convolutional Neural Networks (Mask R-CNNs). In addition, three sets of public images for different disasters were used to test the accuracy of these models. For detecting and marking these two types of structural failures, one of proposed methods can achieve an accuracy of 67.6% and 81.1%, respectively, on low- and high-resolution images collected from field investigations. The results demonstrate that it is feasible to use the proposed end-to-end method for automatically locating and segmenting the damage using 2D images which can help human experts in cases of disasters.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
