David Miller

David Miller


Research Focus & Metrics

Academic Career Length

25Years (2002-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
26 February 2020 in Proceedings of the IEEE
Verified
Journal Article
Closed access
DMDavid Miller
David Miller
ZXZhen Xiang
Zhen Xiang
GKGeorge Kesidis
George Kesidis
With wide deployment of machine learning (ML)-based systems for a variety of applications including medical, military, automotive, genomic, multimedia, and social networking, there is great potential for damage from adversarial learning (AL) attacks. In this article, we provide a contemporary survey of AL, focused particularly on defenses against attacks on deep neural network classifiers. After introducing relevant terminology and the goals and range of possible knowledge of both attackers and defenders, we survey recent work on test-time evasion (TTE), data poisoning (DP), backdoor DP, and reverse engineering (RE) attacks and particularly defenses against the same. In so doing, we distinguish robust classification from anomaly detection (AD), unsupervised from supervised, and statistical hypothesis-based defenses from ones that do not have an explicit null (no attack) hypothesis. We also consider several scenarios for detecting backdoors. We provide a technical assessment for reviewed works, including identifying any issues/limitations, required hyperparameters, needed computational complexity, as well as the performance measures evaluated and the obtained quality. We then delve deeper, providing novel insights that challenge conventional AL wisdom and that target unresolved issues, including: robust classification versus AD as a defense strategy; the belief that attack success increases with attack strength, which ignores susceptibility to AD; small perturbations for TTE attacks: a fallacy or a requirement; validity of the universal assumption that a TTE attacker knows the ground-truth class for the example to be attacked; black, gray, or white-box attacks as the standard for defense evaluation; and susceptibility of query-based RE to an AD defense. We also discuss attacks on the privacy of training data. We then present benchmark comparisons of several defenses against TTE, RE, and backdoor DP attacks on images. The article concludes with a discussion of continuing research directions, including the supreme challenge of detecting attacks whose goal is not to alter classification decisions, but rather simply to embed, without detection, ``fake news'' or other false content.

15 December 2011 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
BZBai Zhang
Bai Zhang
DMDavid Miller
David Miller
YWYue Wang
Yue Wang
Echo state networks (ESNs) are a novel form of recurrent neural networks (RNNs) that provide an efficient and powerful computational model approximating nonlinear dynamical systems. A unique feature of an ESN is that a large number of neurons (the “reservoir”) are used, whose synaptic connections are generated randomly, with only the connections from the reservoir to the output modified by learning. Why a large randomly generated fixed RNN gives such excellent performance in approximating nonlinear systems is still not well understood. In this brief, we apply random matrix theory to examine the properties of random reservoirs in ESNs under different topologies (sparse or fully connected) and connection weights (Bernoulli or Gaussian). We quantify the asymptotic gap between the scaling factor bounds for the necessary and sufficient conditions previously proposed for the echo state property. We then show that the state transition mapping is contractive with high probability when only the necessary condition is satisfied, which corroborates and thus analytically explains the observation that in practice one obtains echo states when the spectral radius of the reservoir weight matrix is smaller than 1.

16 December 2020 in IEEE Transactions on Neural Networks and Learning Systems
Verified
Journal Article
Closed access
ZXZhen Xiang
Zhen Xiang
DMDavid Miller
David Miller
GKGeorge Kesidis
George Kesidis
With wide deployment of deep neural network (DNN) classifiers, there is great potential for harm from adversarial learning attacks. Recently, a special type of data poisoning (DP) attack, known as a backdoor (or Trojan), was proposed. These attacks do not seek to degrade classification accuracy, but rather to have the classifier learn to classify to a target class t* whenever the backdoor pattern is present in a test example originally from a source class s*. Launching backdoor attacks does not require knowledge of the classifier or its training process--only the ability to poison the training set with exemplars containing a backdoor pattern (labeled with the target class). Defenses against backdoors can be deployed before/during training, post-training, or at test time. Here, we address post-training detection in DNN image classifiers, seldom considered in existing works, wherein the defender does not have access to the poisoned training set, but only to the trained classifier itself, as well as to clean (unpoisoned) examples from the classification domain. This scenario is of great interest because e.g., a classifier may be the basis of a phone app that will be shared with many users. Detection may thus reveal a widespread attack. We propose a purely unsupervised anomaly detection (AD) defense against imperceptible backdoor attacks that: 1) detects whether the trained DNN has been backdoor-attacked; 2) infers the source and target classes in a detected attack; 3) estimates the backdoor pattern itself. Our AD approach involves learning (via suitable cost function minimization) the minimum size/norm perturbation (putative backdoor) required to induce the classifier to misclassify (most) examples from class s to class t, for all (s,t) pairs. Our hypothesis is that nonattacked pairs require large perturbations, while the attacked pair (s*, t*) requires much smaller ones. This is convincingly borne out experimentally. We identify a variety of plausible cost functions and devise a novel, robust hypothesis testing approach to perform detection inference. We test our approach, in comparison with the state-of-the-art methods, for several backdoor patterns, attack settings and mechanisms, and data sets and demonstrate its favorability. Our defense essentially requires setting a single hyperparameter (the detection threshold), which can e.g., be chosen to fix the system's false positive rate.\enlargethispage8pt

16 July 2009 in Bioinformatics
Verified
Journal Article
Open access
PDF file
DMDavid Miller
David Miller
YZYanxin Zhang
Yanxin Zhang
GYGuoqiang Yu
Guoqiang Yu
YLYongmei Liu
Yongmei Liu
LCLi Chen
Li Chen
YWYue Wang
Yue Wang
Motivation: In both genome-wide association studies (GWAS) and pathway analysis, the modest sample size relative to the number of genetic markers presents formidable computational, statistical and methodological challenges for accurately identifying markers/interactions and for building phenotype-predictive models. Results: We address these objectives via maximum entropy conditional probability modeling (MECPM), coupled with a novel model structure search. Unlike neural networks and support vector machines (SVMs), MECPM makes explicit and is determined by the interactions that confer phenotype-predictive power. Our method identifies both a marker subset and the multiple k-way interactions between these markers. Additional key aspects are: (i) evaluation of a select subset of up to five-way interactions while retaining relatively low complexity; (ii) flexible single nucleotide polymorphism (SNP) coding (dominant, recessive) within each interaction; (iii) no mathematical interaction form assumed; (iv) model structure and order selection based on the Bayesian Information Criterion, which fairly compares interactions at different orders and automatically sets the experiment-wide significance level; (v) MECPM directly yields a phenotype-predictive model. MECPM was compared with a panel of methods on datasets with up to 1000 SNPs and up to eight embedded penetrance function (i.e. ground-truth) interactions, including a five-way, involving less than 20 SNPs. MECPM achieved improved sensitivity and specificity for detecting both ground-truth markers and interactions, compared with previous methods. Availability:http://www.cbil.ece.vt.edu/ResearchOngoingSNP.htm Contact:djmiller@engr.psu.edu Supplementary information: Supplementary data are available at Bioinformatics online.

1 October 2013 in Osteoarthritis and Cartilage
Verified
Journal Article
Closed access
KUK.L. Urish
K.L. Urish
MKM.G. Keffalas
M.G. Keffalas
JDJ.R. Durkin
J.R. Durkin
DMDavid Miller
David Miller
CCC.R. Chu
C.R. Chu
TMT.J. Mosher
T.J. Mosher

25 February 2010 in IEEE Transactions on Neural Networks
Verified
Journal Article
Closed access
YAYaman Aksu
Yaman Aksu
DMDavid Miller
David Miller
GKGeorge Kesidis
George Kesidis
QYQing X. Yang
Qing X. Yang
Feature selection for classification in high-dimensional spaces can improve generalization, reduce classifier complexity, and identify important, discriminating feature ¿markers.¿ For support vector machine (SVM) classification, a widely used technique is recursive feature elimination (RFE). We demonstrate that RFE is not consistent with margin maximization, central to the SVM learning approach. We thus propose explicit margin-based feature elimination (MFE) for SVMs and demonstrate both improved margin and improved generalization, compared with RFE. Moreover, for the case of a nonlinear kernel, we show that RFE assumes that the squared weight vector 2-norm is strictly decreasing as features are eliminated. We demonstrate this is not true for the Gaussian kernel and, consequently, RFE may give poor results in this case. MFE for nonlinear kernels gives better margin and generalization. We also present an extension which achieves further margin gains, by optimizing only two degrees of freedom-the hyperplane's intercept and its squared 2-norm-with the weight vector orientation fixed. We finally introduce an extension that allows margin slackness. We compare against several alternatives, including RFE and a linear programming method that embeds feature selection within the classifier design. On high-dimensional gene microarray data sets, University of California at Irvine (UCI) repository data sets, and Alzheimer's disease brain image data, MFE methods give promising results.