Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Adversarial examples”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4

Multilayered Network Models for Security: Enhancing System Security Engineering with Orchestration

Security engineering approaches can often focus on a particular domain—physical security, cyber security, or personnel security, for example. Yet, security systems engineering consistently faces challenges requiring socio-technical solutions to address evolving and dynamic complexity. While some drivers of this complexity stem from complex risk environments, innovative adversaries, and disruptive technologies, other drivers are endogenous and emerge from the interactions across security engineering approaches. In response, INCOSE's Systems Security Working Group identified the need to better coordinate “disparate security solutions [that] operate independently” as one of eleven key concepts in their IS21 FuSE Security Roadmap. From this perspective, this need for “security orchestration” aligns with the perspective that security is a property that emerges from interactions within complex systems. Current efforts at Sandia National Laboratories are developing a systems security engineering approach that describes high consequence facility (HCF) security as a multidomain set of interacting layers. The result is a multilayered network (MLN)-based approach that captures the interactions between infrastructure, physical components, digital components, and humans in nuclear security systems. This article will summarize the MLN-based approach to HCF security and describe two preliminary results demonstrating potential benefits from incorporating interactions across disparate security solutions. Here, leveraging the logical structure of networks, this MLN model-based approach provides an example of how security orchestration provides enhanced systems security engineering solutions.

42 ENGINEERING↗

Proactive Intrusion Detection and Mitigation System

SAND2023-05661O The proactive intrusion detection and mitigation system (PIDMS) provides grid-edge situational awareness for cybersecurity defense by capturing real-time distributed energy resource (DER) network traffic and performance data with a novel approach that improves the detection and prevention of cyber-physical attacks. The PIDMS addresses the grid-edge security gap with real-time analysis of both network traffic and photovoltaic performance data to deliver a novel, cyber-physical intrusion detection system (IDS) approach that increases the accuracy and effectiveness of detection and mitigation. This hybrid IDS analysis enables dual monitoring that increases the workload of the adversary; both cyber and physical data would have to be simultaneously spoofed to evade detection. Furthermore, monitoring and analyzing cyber data are insufficient in some cases. For example, in an insider threat aimed at disrupting inverter grid-support functions where proper credentials and authentication are achieved, only the altered PV performance would indicate abnormal behavior. All in all, the PIDMS provides novel capabilities for: • Distributed, real-time cyber-physical detection and mitigation analysis • Cybersecurity defense for grid-edge systems • Analysis framework that can provide situational awareness across the transmission, distribution, and DER systems The PIDMS sensor is designed to collect cyber-physical data, process the data using machine-learning algorithms, detect abnormal events, and deploy mitigations. With these goals, the main functional PIDMS objectives are: • Capability to collect cyber-physical data • Onboard storage of cyber-physical data • Peer-to-peer communication • Computationally efficient machine-learning algorithms • Online cyber-physical data analysis • Alerting/visualization capabilities • Mitigation deployment capability with bump-in-the-wire (BITW) implementation Each of these functional objectives enable PIDMS to perform effective cyber-physical intrusion detection and mitigation. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Jones, Christian↗

Robust Network Interdiction Considering Multiple Defense Resources

We extend the network interdiction problem in order to consider multiple options for defending arcs in a network. We apply our model to select an optimal set of countermeasures for nuclear material smuggling. In order account for the joint impact of multiple countermeasures, we consider multiple components of defense: situational awareness, detection, and interdiction. We develop a bi-level programming model, where the adversary's goal is to maximize damage to the system while the defender seeks to minimize damage by selecting countermeasure options under a restricted bud- get. Furthermore, since effectiveness of countermeasures is often difficult to assess with high accuracy, we use robust optimization to allow for parametric uncertainty. We test our method on a synthetic example of the U.S. supply chain network that includes rail, air, and maritime transportation between both domestic and international locations. For this test example, we find that increasing the budget for countermeasures has the greatest impact within the first 1-2% of the cost to implement all countermeasures and that the uncertainty tolerance has a greater impact on expected damage than the countermeasure budget does. Thus, when strategically selecting countermeasures it is important to precisely understand the effectiveness of each option and the adversary's probability of subverting them.

97 MATHEMATICS AND COMPUTING↗

PUF Entropy

This note discusses aspects of fuzzy extraction: in particular, how the interplay between PUF entropy and the revelation of helper data affects the entropy of the underlying secret. We examine the theory behind a scheme by Dodis et alii to develop exact formulas for the entropy of the given seed, given that an adversary knows the helper data. We also give a simple lower bound on the security of the Dodis scheme that is achieved exactly in many instances. By way of several examples, we quantify the amount of entropy that can be lost due to interactions between imperfect PUF implementations and given error correction methods. We repeatedly show that understanding the details of the error correction scheme and the interaction with the PUF response is crucial to being able to estimate the entropy of the system.

97 MATHEMATICS AND COMPUTING↗

Decoding structure-spectrum relationships with physically organized latent spaces

Here, a semisupervised machine learning method for the discovery of structure-spectrum relationships is developed and then demonstrated using the specific example of interpreting x-ray absorption near-edge structure (XANES) spectra. This method constructs a one-to-one mapping between individual structure descriptors and spectral trends. Specifically, an adversarial autoencoder is augmented with a rank constraint (RankAAE). The RankAAE methodology produces a continuous and interpretable latent space, where each dimension can track an individual structure descriptor. As a part of this process, the model provides a robust and quantitative measure of the structure-spectrum relationship by decoupling intertwined spectral contributions from multiple structural characteristics. This makes it ideal for spectral interpretation and the discovery of descriptors. The capability of this procedure is showcased by considering five local structure descriptors and a database of >50 000 simulated XANES spectra across eight first-row transition metal oxide families. The resulting structure-spectrum relationships not only reproduce known trends in the literature but also reveal unintuitive ones that are visually indiscernible in large datasets. The results suggest that the RankAAE methodology has great potential to assist researchers in interpreting complex scientific data, testing physical hypotheses, and revealing patterns that extend scientific insight.

36 MATERIALS SCIENCE↗

Adversarial methods to reduce simulation bias in neutrino interaction event filtering at liquid argon time projection chambers

For current and future neutrino oscillation experiments using large liquid argon time projection chambers (LAr-TPCs), a key challenge is identifying neutrino interactions from the pervading cosmic-ray background. Rejection of such background is often possible using traditional cut-based selections, but this typically requires the prior use of computationally expensive reconstruction algorithms. This work demonstrates an alternative approach of using a 3D submanifold sparse convolutional network trained on low-level information from the scintillation light signal of interactions inside LAr-TPCs. This technique is applied to example simulations from ICARUS, the far detector of the short baseline neutrino program at Fermilab. The results of the network, show that cosmic background is reduced by up to 76.3% whilst neutrino interaction selection efficiency remains over 98.9%. We further present a way to mitigate potential biases from imperfect input simulations by applying domain adversarial neural networks (DANNs), for which modified simulated samples are introduced to imitate real data and a small portion of them are used for adversarial training. A series of mock-data studies are performed and demonstrate the effectiveness of using DANNs to mitigate biases, showing neutrino interaction selection efficiency performances significantly better than that achieved without the adversarial training.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Towards Resilient Design of Leader-following Consensus with Attack Identification and Privacy Preservation Capabilities

This paper considers a leader-following consensus in the presence of unknown but bounded cyber-attacks. Specifically, we consider the following cyber-attack scenarios: (i) an attacker aims to destabilize the consensus dynamics by injecting exogenous signals to both the actuators of the followers and/or the communication network, (ii) an eavesdropper adversary aims to obtain information on the physical state of the agents. To this end, a novel resilient leader-following consensus algorithm based on a competitive interaction method is proposed. In addition, it is demonstrated that by appropriately choosing the information exchanged between the agents, the proposed control framework also enables the cooperative system to either distributively identify the compromised communication links in real-time or to protect the privacy of the physical state of the agents from the eavesdropper. Here, a numerical example is provided to illustrate the proposed resilient control algorithms.

Gusrialdi, Azwirman↗

Resource cost results for one-way entanglement distillation and state merging of compound and arbitrarily varying quantum sources

We consider one-way quantum state merging and entanglement distillation under compound and arbitrarily varying source models. Regarding quantum compound sources, where the source is memoryless, but the source state an unknown member of a certain set of density matrices, we continue investigations begun in the work of Bjelaković et al. [“Universal quantum state merging,” J. Math. Phys. 54, 032204 (2013)] and determine the classical as well as entanglement cost of state merging. We further investigate quantum state merging and entanglement distillation protocols for arbitrarily varying quantum sources (AVQS). In the AVQS model, the source state is assumed to vary in an arbitrary manner for each source output due to environmental fluctuations or adversarial manipulation. We determine the one-way entanglement distillation capacity for AVQS, where we invoke the famous robustification and elimination techniques introduced by Ahlswede. Regarding quantum state merging for AVQS we show by example that the robustification and elimination based approach generally leads to suboptimal entanglement as well as classical communication rates.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Improving Deep Neural Networks’ Training for Image Classification With Nonlinear Conjugate Gradient-Style Adaptive Momentum

Momentum is crucial in stochastic gradient-based optimization algorithms for accelerating or improving training deep neural networks (DNNs). In deep learning practice, the momentum is usually weighted by a well-calibrated constant. However, tuning the hyperparameter for momentum can be a significant computational burden. In this article, we propose a novel adaptive momentum for improving DNNs training; this adaptive momentum, with no momentum-related hyperparame- ter required, is motivated by the nonlinear conjugate gradient (NCG) method. Stochastic gradient descent (SGD) with this new adaptive momentum eliminates the need for the momentum hyperparameter calibration, allows using a significantly larger learning rate, accelerates DNN training, and improves the final accuracy and robustness of the trained DNNs. For example, SGD with this adaptive momentum reduces classification errors for training ResNet110 for CIFAR10 and CIFAR100 from 5.25% to 4.64% and 23.75% to 20.03%, respectively. Furthermore, SGD, with the new adaptive momentum, also benefits adversarial training and, hence, improves the adversarial robustness of the trained DNNs.

97 MATHEMATICS AND COMPUTING↗

Miscalculation, Misperception and Risk Reduction

In the current security environment, the most likely wars for which the US and its allies must be prepared are wars against nuclear-armed states. Such wars bring unique risks of both vertical and horizontal escalation. Some are “accepted risks;” for example, the US may choose to escalate a conflict to clarify its commitment to an issue and defend its interest. Other are “unacceptable risks;” especially, the risk of miscalculation by the U.S. or an adversary. The focus here is on this latter category. Miscalculation is an error of judgment—the failure to accurately assess a situation. Miscalculation can result from hidden, incomplete, or ambiguous information, or from inaccurate processing of available information because of bias, overconfidence, denial, delusion, etc. Miscalculation is related to misperception. For some, they are synonymous. This paper characterizes the risks of miscalculation and introduces a discussion of a risk mitigation strategy.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Adversarial uncertainty quantification in physics-informed neural networks

Here, we introduce a deep learning framework for quantifying and propagating uncertainty in systems governed by non-linear differential equations using physics-informed neural networks. Specifically, we employ latent variable models to construct probabilistic representations for the system states, and put forth an adversarial inference procedure for training them on data, while constraining their predictions to satisfy given physical laws expressed by partial differential equations. Such physics-informed constraints provide a regularization mechanism for effectively training deep generative models as surrogates of physical systems in which the cost of data acquisition is high, and training data-sets are typically small. This provides a flexible framework for characterizing uncertainty in the outputs of physical systems due to randomness in their inputs or noise in their observations that entirely bypasses the need for repeatedly sampling expensive experiments or numerical simulators. Moreover, we demonstrate the effectiveness of our approach through a series of examples involving uncertainty propagation in non-linear conservation laws, and the discovery of constitutive laws for flow through porous media directly from noisy data.

97 MATHEMATICS AND COMPUTING↗

Assessment of Russian VSTOL technology evaluating the YAK-38 'FORGER' and YAK-141 'FREESTYLE'

The dissolution of the Former Soviet Union (FSU) created new relationships between the world superpowers. Overnight, the Commonwealth of Independent States (CIS), formed from the remnants of the FSU, began the difficult transformation to a free market society. Military hardware that had once been highly classified and the basis for our own defense planning was now openly marketed at airshows around the world. 'Test' flights were available for potential customers and cooperative partnerships were explored between former adversaries. This environment permitted a visit to the Yakovlev Design Bureau, (YAK) for a vertical/short takeoff and landing (VSTOL) technology assessment. Yakovlev is the FSU's sole Design Bureau with experience in VSTOL aircraft and has developed two flying examples, the YAK-38 'FORGER' and YAK-141 'FREESTYLE'. This article reviews the performance of the YAK-38 'FORGER' and the YAK-141 'FREESTYLE'.

Nalls, Art↗

Detecting CAN Masquerade Attacks with Signal Clustering Similarity

Vehicular Controller Area Networks (CANs) are susceptible to cyber attacks of different levels of sophistication. Fabrication attacks are the easiest to administer—an adversary simply sends (extra) frames on a CAN—but also the easiest to detect because they disrupt frame frequency. To overcome time-based detection methods, adversaries must administer masquerade attacks by sending frames in lieu of (and therefore at the expected time of) benign frames but with malicious payloads. Research efforts have proven that CAN attacks, and masquerade attacks in particular, can affect vehicle functionality. Examples include causing unintended acceleration, deactivation of vehicle’s brakes, as well as steering the vehicle. We hypothesize that masquerade attacks modify the nuanced correlations of CAN signal time series and how they cluster together. Therefore, changes in cluster assignments should indicate anomalous behavior. We confirm this hypothesis by leveraging our previously developed capability for reverse engineering CAN signals (i.e., CAN-D [Controller Area Network Decoder]) and focus on advancing the state of the art for detecting masquerade attacks by analyzing time series extracted from raw CAN frames. Specifically, we demonstrate that masquerade attacks can be detected by computing time series clustering similarity using hierarchical clustering on the vehicle’s CAN signals (time series) and comparing the clustering similarity across CAN captures with and without attacks. We test our approach in a previously collected CAN dataset with masquerade attacks (i.e., the ROAD dataset) and develop a forensic tool as a proof of concept to demonstrate the potential of the proposed approach for detecting CAN masquerade attacks.

Moriano Salazar, Pablo↗

Space in 21st Century Conflict: Calibrating Risks, Tailoring Strategies (Workshop Summary)

U.S. strategic thinking on outer space security has been slow to evolve since the 1990s and the DoD finds itself struggling to develop an adaptive and cohesive strategy for the domain. Russia and China, however, have been thinking about space in a holistic and integrated way for some time. Both countries recognize that U.S. conventional strength relies heavily on technologies and systems enabled by space-based assets. Thus, holding those assets at risk represents an opportunity to asymmetrically counter America’s ability to project power. China and Russia been developing a suite of anti-satellite (ASAT) capabilities to accomplish this, to include direct-ascent weapons, space-based weapons, jammers, and cyber capabilities; both countries have also incorporated these systems into their respective tactical and strategic planning. This represents a serious threat to U.S. power projection and assurance of our regional allies and partners. Assuring the U.S.’s ability to operate in space is paramount for supporting regional security missions in both Europe and East Asia. Assuring DoD operations and capabilities in outer space, however, remains a challenge for the defense community. One solution offered by strategists is to enhance U.S. resiliency in space to turn the domain from offense dominant to defense dominant. This, they argue, can be achieved by improving space situational awareness (SSA), protecting space assets with active and passive countermeasures, disaggregating assets and leveraging new commercial space capabilities, and integrating allied space capabilities. However, many barriers remain to accomplishing these objectives. For example, some think that the classified nature of many U.S. space activities is an obstacle to deepening cooperation with commercial partners and allies. Others acknowledge the lack of a working model for incorporating cheaper, diverse, and scalable commercial space systems. Identifying where space fits into broader U.S. deterrence strategies is a subject of contention. While discussions over the past few years have focused on the practicability of deterring attacks on space assets via cross-domain threats, this line of thinking has yielded few useful insights. Consequently, defense practitioners have more recently shifted to thinking of space as part of an integrated deterrence concept, one that envisions building resiliency and defenses into our space architecture while simultaneously holding an adversary’s space-based assets at risk to enhance deterrence more broadly. However, both approaches illuminate a larger problem in the discussion of the relationship between space and deterrence: there is a lack of consensus on what it is we are trying to deter, how space fits into broader deterrence and strategic constructs, and the role of space in extended deterrence and assurance of allies. For example, is the United States’ priority to discourage attacks in space, to ensure regional power projection and alliance support capabilities, to deter broader global conflict or all of the above? Answering such questions will be an important step toward integrating space into a comprehensive defense strategy.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗

End-to-end learning of multiple sequence alignments with differentiable Smith–Waterman

Abstract Motivation Multiple sequence alignments (MSAs) of homologous sequences contain information on structural and functional constraints and their evolutionary histories. Despite their importance for many downstream tasks, such as structure prediction, MSA generation is often treated as a separate pre-processing step, without any guidance from the application it will be used for. Results Here, we implement a smooth and differentiable version of the Smith–Waterman pairwise alignment algorithm that enables jointly learning an MSA and a downstream machine learning system in an end-to-end fashion. To demonstrate its utility, we introduce SMURF (Smooth Markov Unaligned Random Field), a new method that jointly learns an alignment and the parameters of a Markov Random Field for unsupervised contact prediction. We find that SMURF learns MSAs that mildly improve contact prediction on a diverse set of protein and RNA families. As a proof of concept, we demonstrate that by connecting our differentiable alignment module to AlphaFold2 and maximizing predicted confidence, we can learn MSAs that improve structure predictions over the initial MSAs. Interestingly, the alignments that improve AlphaFold predictions are self-inconsistent and can be viewed as adversarial. This work highlights the potential of differentiable dynamic programming to improve neural network pipelines that rely on an alignment and the potential dangers of optimizing predictions of protein sequences with methods that are not fully understood. Availability and implementation Our code and examples are available at: https://github.com/spetti/SMURF. Supplementary information Supplementary data are available at Bioinformatics online.

59 BASIC BIOLOGICAL SCIENCES↗

Making Corgis Important for Honeycomb Classification: Adversarial Attacks on Concept-based Explainability Tools

Methods for model explainability have become increasingly critical for testing the fairness and soundness of deep learning. Concept-based interpretability techniques, which use a small set of human-interpretable concept exemplars in order to measure the influence of a concept on a model's internal representation of input, are an important thread in this line of research. In this work we show that these explainability methods can suffer the same vulnerability to adversarial attacks as the models they are meant to analyze. We demonstrate this phenomenon on two well-known concept-based interpretability methods: TCAV and faceted feature visualization. We show that by leveraging the geometry of the problem and carefully perturbing the examples of the concept that is being investigated, we can radically change the output of the interpretability method. The attacks that we propose can either induce positive interpretations (polka dots are an important concept for a model when classifying zebras) or negative interpretations (stripes are not an important factor in identifying images of a zebra). Our work highlights the fact that in safety-critical applications, there is need for security around not only the machine learning pipeline but also the model interpretation process.

Brown, Davis R.↗

DeepMerge – II. Building robust deep learning algorithms for merging galaxy identification across domains

In astronomy, neural networks are often trained on simulation data with the prospect of being used on telescope observations. Unfortunately, training a model on simulation data and then applying it to instrument data leads to a substantial and potentially even detrimental decrease in model accuracy on the new target dataset. Simulated and instrument data represent different data domains, and for an algorithm to work in both, domain-invariant learning is necessary. Here we employ domain adaptation techniques— Maximum Mean Discrepancy (MMD) as an additional transfer loss and Domain Adversarial Neural Networks (DANNs)— and demonstrate their viability to extract domain-invariant features within the astronomical context of classifying merging and non-merging galaxies. Additionally, we explore the use of Fisher loss and entropy minimization to enforce better in-domain class discriminability. We show that the addition of each domain adaptation technique improves the performance of a classifier when compared to conventional deep learning algorithms. We demonstrate this on two examples: between two Illustris-1 simulated datasets of distant merging galaxies, and between Illustris-1 simulated data of nearby merging galaxies and observed data from the Sloan Digital Sky Survey. The use of domain adaptation techniques in our experiments leads to an increase of target domain classification accuracy of up to ~20%. With further development, these techniques will allow astronomers to successfully implement neural network models trained on simulation data to efficiently detect and study astrophysical objects in current and future large-scale astronomical surveys.

galaxies: interactions↗

A Graph-Based Impact Metric for Mitigating Lateral Movement Cyber Attacks

Most cyber network attacks begin with an adversary gain- ing a foothold within the network and proceed with lateral movement until a desired goal is achieved. The mechanism by which lateral movement occurs varies but the basic signa- ture of hopping between hosts by exploiting vulnerabilities is the same. Because of the nature of the vulnerabilities typ- ically exploited, lateral movement is very difficult to detect and defend against. In this paper we define a dynamic reach- ability graph model of the network to discover possible paths that an adversary could take using different vulnerabilities, and how those paths evolve over time. We use this reacha- bility graph to develop dynamic machine-level and network- level impact scores. Lateral movement mitigation strategies which make use of our impact scores are also discussed, and we detail an example using a freely available data set.

cyber security↗