Physical Sciences and Mathematics Commons^™

Stealing Deep Reinforcement Learning Models For Fun And Profit, Kangjie Chen, Shangwei Guo, Tianwei Zhang, Xiaofei Xie, Yang Liu

Research Collection School Of Computing and Information Systems

This paper presents the first model extraction attack against Deep Reinforcement Learning (DRL), which enables an external adversary to precisely recover a black-box DRL model only from its interaction with the environment. Model extraction attacks against supervised Deep Learning models have been widely studied. However, those techniques cannot be applied to the reinforcement learning scenario due to DRL models' high complexity, stochasticity and limited observable information. We propose a novel methodology to overcome the above challenges. The key insight of our approach is that the process of DRL model extraction is equivalent to imitation learning, a well-established solution to learn …

Order-Agnostic Cross Entropy For Non-Autoregressive Machine Translation, Cunxiao Du, Zhaopeng Tu, Jing Jiang

Research Collection School Of Computing and Information Systems

We propose a new training objective named orderagnostic cross entropy (OAXE) for fully nonautoregressive translation (NAT) models. OAXE improves the standard cross-entropy loss to ameliorate the effect of word reordering, which is a common source of the critical multimodality problem in NAT. Concretely, OAXE removes the penalty for word order errors, and computes the cross entropy loss based on the best possible alignment between model predictions and target tokens. Since the log loss is very sensitive to invalid references, we leverage cross entropy initialization and loss truncation to ensure the model focuses on a good part of the search space. …

A Coprocessor-Based Introspection Framework Via Intel Management Engine, Lei Zhou, Fengwei Zhang, Jidong Xiao, Kevin Leach, Westley Weimer, Xuhua Ding, Guojun Wang

Research Collection School Of Computing and Information Systems

During the past decade, virtualization-based (e.g., virtual machine introspection) and hardware-assisted approaches (e.g., x86 SMM and ARM TrustZone) have been used to defend against low-level malware such as rootkits. However, these approaches either require a large Trusted Computing Base (TCB) or they must share CPU time with the operating system, disrupting normal execution. In this article, we propose an introspection framework called NIGHTHAWK that transparently checks system integrity and monitor the runtime state of target system. NIGHTHAWK leverages the Intel Management Engine (IME), a co-processor that runs in isolation from the main CPU. By using the IME, our approach has …

A Mean-Field Markov Decision Process Model For Spatial-Temporal Subsidies In Ride-Sourcing Markets, Zheng Zhu, Jintao Ke, Hai Wang

Research Collection School Of Computing and Information Systems

Ride-sourcing services are increasingly popular because of their ability to accommodate on-demand travel needs. A critical issue faced by ride-sourcing platforms is the supply-demand imbalance, as a result of which drivers may spend substantial time on idle cruising and picking up remote passengers. Some platforms attempt to mitigate the imbalance by providing relocation guidance for idle drivers who may have their own self-relocation strategies and decline to follow the suggestions. Platforms then seek to induce drivers to system-desirable locations by offering them subsidies. This paper proposes a mean-field Markov decision process (MF-MDP) model to depict the dynamics in ride-sourcing markets …

Vibransee: Enabling Simultaneous Visible Light Communication And Sensing, Ila Nitin Gokarn, Archan Misra

Research Collection School Of Computing and Information Systems

Driven by the ubiquitous proliferation of low-cost LED luminaires, visible light communication (VLC) has been established as a high-speed communications technology based on the high-frequency modulation of an optical source. In parallel, Visible Light Sensing (VLS) has recently demonstrated how vision-based at-a-distance sensing of mechanical vibrations (e.g., of factory equipment) can be performed using high frequency optical strobing. However, to date, exemplars of VLC and VLS have been explored in isolation, without consideration of their mutual dependencies. In this work, we explore whether and how high-throughput VLC and high-coverage VLS can be simultaneously supported. We first demonstrate the existence of …

Efficient White-Box Fairness Testing Through Gradient Search, Lingfeng Zhang, Yueling Zhang, Min Zhang

Research Collection School Of Computing and Information Systems

Deep learning (DL) systems are increasingly deployed for autonomous decision-making in a wide range of applications. Apart from the robustness and safety, fairness is also an important property that a well-designed DL system should have. To evaluate and improve individual fairness of a model, systematic test case generation for identifying individual discriminatory instances in the input space is essential. In this paper, we propose a framework EIDIG for efficiently discovering individual fairness violation. Our technique combines a global generation phase for rapidly generating a set of diverse discriminatory seeds with a local generation phase for generating as many individual discriminatory …

Task Similarity Aware Meta Learning: Theory-Inspired Improvement On Maml, Pan Zhou, Yingtian Zpu, Xiaotong Yuan, Jiashi Feng, Caiming Xiong, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

Few-shot learning ability is heavily desired for machine intelligence. By meta-learning a model initialization from training tasks with fast adaptation ability to new tasks, model-agnostic meta-learning (MAML) has achieved remarkable success in a number of few-shot learning applications. However, theoretical understandings on the learning ability of MAML remain absent yet, hindering developing new and more advanced meta learning methods in a principled way. In this work, we solve this problem by theoretically justifying the fast adaptation capability of MAML when applied to new tasks. Specifically, we prove that the learnt meta-initialization can benefit the fast adaptation to new tasks with …

Functionalized Ionic Liquids For Improved Co2 Hydrogenation Catalysts, Amol Nanduji Jaybhaye

Masters Theses & Specialist Projects

Climate change is one of the biggest problems we are facing today, and anthropogenic greenhouse gases contribute significantly to it. Among greenhouse gasses, carbon dioxide emission is increased drastically and affecting majorly for increasing global temperature, and anything we can do to mitigate carbon dioxide emission will be helpful. A carbon dioxide hydrogenation reaction is a promising technique that can convert carbon dioxide into value-added chemicals, but the reaction rate is slow, and for that, we can use catalysts that will accelerate the reaction rate. Platinum-based nanoparticles have sparked research related to energy and environmental catalysts. Catalytic properties are depending …

Optimization Planning For 3d Convnets, Zhaofan Qiu, Ting Yao, Chong-Wah Ngo, Tao Mei

Research Collection School Of Computing and Information Systems

It is not trivial to optimally learn a 3D Convolutional Neural Networks (3D ConvNets) due to high complexity and various options of the training scheme. The most common hand-tuning process starts from learning 3D ConvNets using short video clips and then is followed by learning long-term temporal dependency using lengthy clips, while gradually decaying the learning rate from high to low as training progresses. The fact that such process comes along with several heuristic settings motivates the study to seek an optimal "path" to automate the entire training. In this paper, we decompose the path into a series of training …

Frameaxis: Characterizing Microframe Bias And Intensity With Word Embedding, Haewoon Kwak, Jisun An, Elise Jing Jing, Yong-Yeol Ahn

Research Collection School Of Computing and Information Systems

Framing is a process of emphasizing a certain aspect of an issue over the others, nudging readers or listeners towards different positions on the issue even without making a biased argument. Here, we propose FrameAxis, a method for characterizing documents by identifying the most relevant semantic axes (“microframes”) that are overrepresented in the text using word embedding. Our unsupervised approach can be readily applied to large datasets because it does not require manual annotations. It can also provide nuanced insights by considering a rich set of semantic axes. FrameAxis is designed to quantitatively tease out two important dimensions of how …

Self-Supervised Contrastive Learning For Code Retrieval And Summarization Via Semantic-Preserving Transformations, Duy Quoc Nghi Bui, Yijun Yu, Lingxiao Jiang

Research Collection School Of Computing and Information Systems

We propose Corder, a self-supervised contrastive learning framework for source code model. Corder is designed to alleviate the need of labeled data for code retrieval and code summarization tasks. The pre-trained model of Corder can be used in two ways: (1) it can produce vector representation of code which can be applied to code retrieval tasks that do not have labeled data; (2) it can be used in a fine-tuning process for tasks that might still require label data such as code summarization. The key innovation is that we train the source code model by asking it to recognize similar …

Claim: Curriculum Learning Policy For Influence Maximization In Unknown Social Networks, Dexun Li, Meghna Lowalekar, Pradeep Varakantham

Research Collection School Of Computing and Information Systems

Influence maximization is the problem of finding a small subset of nodes in a network that can maximize the diffusion of information. Recently, it has also found application in HIV prevention, substance abuse prevention, micro-finance adoption, etc., where the goal is to identify the set of peer leaders in a real-world physical social network who can disseminate information to a large group of people. Unlike online social networks, real-world networks are not completely known, and collecting information about the network is costly as it involves surveying multiple people. In this paper, we focus on this problem of network discovery for …

Polyaromatic-Terminated Iron Polypyridyl Complexes For The Functionalization Of Carbon Surfaces And Electrocatalytic Hydrogen Generation, Caroline Marie Margonis

Dissertations, Theses, and Masters Projects

Artificial Photosynthesis (AP) focuses on developing methods for the conversion of solar energy into chemical fuel in the form of H2 and O2. Heterogeneous photocatalytic systems incorporating carbon nanotubes (CNTs) have shown much promise but are currently limited and expensive due to their reliance on noble metals. To that end, this work focuses on the development and synthesis of cheaper naphthalene- and pyrene-terminated iron polypyridyl complexes for use in the simultaneous functionalization of carbon surfaces, electrocatalytic proton reduction, and eventual incorporation in photocatalytic systems. Cyclic voltammetry was used to characterize the adsorption behavior of each complex on the surface of …

Water Clarity And Suspended Particle Dynamics In The Chesapeake Bay: Local Effects Of Oyster Aquaculture, Regional Effects Of Reduced Shoreline Erosion, And Long-Term Trends In Remotely Sensed Reflectance, Jessica Turner

Dissertations, Theses, and Masters Projects

Water clarity is a key indicator of the ecosystem health in the Chesapeake Bay. Estuarine water clarity fluctuates due to external inputs from the watershed as well as processes occurring within the estuary itself, such as sediment resuspension and organic matter production. Therefore, water clarity requires study at multiple spatial and temporal scales and with multiple metrics. One local-scale process potentially influencing water clarity is shellfish aquaculture. One part of this dissertation examined how water quality and hydrodynamics varied among oyster farms as well as inside versus outside the extent of caged areas located in southern Chesapeake Bay. Current speed …

Cellular Remains In A ~3.42-Billion-Year-Old Subseafloor Hydrothermal Environment, Barbara Cavalazzi, Laurence Lamelle, Alexandre Simionovici, Sherry L. Cady, Michael J. Russell, Elena Bailo, Roberto Canteri, Emanuele Enrico, Alain Manceau, Multiple Additional Authors

Geology Faculty Publications and Presentations

Subsurface habitats on Earth host an extensive extant biosphere and likely provided one of Earth’s earliest microbial habitats. Although the site of life’s emergence continues to be debated, evidence of early life provides insights into its early evolution and metabolic affinity. Here, we present the discovery of exceptionally well-preserved, ~3.42-billion-year-old putative filamentous microfossils that inhabited a paleo-subseafloor hydrothermal vein system of the Barberton greenstone belt in South Africa. The filaments colonized the walls of conduits created by low-temperature hydrothermal fluid. Combined with their morphological and chemical characteristics as investigated over a range of scales, they can be considered the oldest …

Locating Identities In Time: An Examination Of The Impact Of Temporality On Presentations Of The Self Through Location-Based Social Networks, Konstantinos Papangelis, Ioanna Lykourentzou, Vassilis-Javed Khan, Alan Chamberlain, Ting Cao, Micahel Saker, Nicolas Lalone

Articles

Studies of identity and location-based social networks (LBSN) have tended to focus on the performative aspects associated with marking one’s location. Yet, these studies often present this practice as being an a priori aspect of locative media. What is missing from this research is a more granular understanding of how this process develops over time. Accordingly, we focus on the first six weeks of 42 users beginning to use an LBSN we designed and named GeoMoments. Through our analysis of our users' activities, we contribute to understanding identity and LBSN in two distinct ways. First, we show how LBSN users …

A Look Into Increasing The Number Of Veterans And Former Government Employees Converting To Career And Technical Cybersecurity Teachers, Vukica M. Jovanovic, Michael Anthony Crespo, Drew E. Brown, Deborah Marshall, Otilia Popescu, Murat Kuzlu, Petros J. Katsioloudis, Linda Vahala

Engineering Technology Faculty Publications

The current state of technology with recent explosions in the digital processing of paperwork, computer networking use, and online and virtual approaches to areas, which until very recently had traditional and non-computerized ways of operating, led to a steady increase in the demand for jobs in the area of computer science and cybersecurity. The education system, the pipeline for the incoming workforce, needs to keep up with this tremendous pace in technology and the job market. The current K-12 school system has been extensively challenged to fill out necessary positions in order to address the increasing need for programs that …

Characterization Of Ph – Responsive Nanocage Based On The Ferritin Iron Storage Protein, Satyam Singh

Theses and Dissertations

The iron-storage protein ferritin (Ftn) assembles into a protein cage structure with 24 subunits and octahedral (4-fold, 3-fold, 2-fold) symmetry. Each monomeric subunit contains a robust four-helix bundle fold. The fully assembled Ftn structure has a high degree of thermal stability (up to 100°C), a mono dispersed size (12 nm in diameter), and a large central cavity (7-8 nm in diameter). The central cavity stores ferric iron in phylogenetically diverse group of organisms, including humans. The central cavity has been used for encapsulation of cargoes such as other metals, contrast agents for imaging, small molecule drugs for therapy, …

Stemloop-Finder: A Tool For The Detection Of Dna Hairpins With Conserved Motifs., Alyssa A. Pratt, Ellis L. Torrance, George W. Kasun, Kenneth M. Stedman, Ignacio De La Higuera

Center for Life in Extreme Environments Publications

Nucleic acid secondary structures play important roles in regulating biological processes. StemLoop-Finder is a computational tool to recognize and annotate conserved structural motifs in large data sets. The program is optimized for the detection of stem-loop structures that may serve as origins of replication in circular replication-associated protein (Rep)-encoding single-stranded (CRESS) DNA viruses.

J/Ψ Photoproduction Near Threshold With Clas12, Joseph Newton

Physics Theses & Dissertations

The structure of the proton is comprised of quarks and a sea of gluons. A mechanism that can extract the characteristics of the hidden-color correlations of the nuclear wavefunction is the production of charm near threshold. Due to the fact that momentum transfer is large near threshold in the production of J/ψ , all three valence quarks must act coherently to ex- change energy for the reaction to occur. Models have been developed to predict the nature of J/ψ photoproduction at these specific energies. These include production mechanisms with the two-gluon and three-gluon exchanges. The transferred momentum dependence of the …

The Development Of Novel Carbohydrate-Based Gelators And Their Applications As Advanced Soft Materials, Joedian Morris

Chemistry & Biochemistry Theses & Dissertations

Low molecular weight gelators (LMWGs) are attractive molecules that have been explored extensively due to their practical applications in many disciplines. These small molecules self-assemble forming solid-like gels via three-dimensional cross-linked networks with the solvent as the key component within the matrix. Carbohydrate-based LMWGs are small molecules that can form solid-like gels in water, organic solvents, and aqueous solutions. They have great potential to be utilized in different applications because carbohydrates are biocompatible and can be made from easily accessible and renewable resources. Designing gelators is still a challenge within the field, even though researchers have developed tools to predict …

On The Use Of Minimum Penalties In Statistical Learning, Ben Sherwood, Bradley S. Price

Faculty & Staff Scholarship

Modern multivariate machine learning and statistical methodologies estimate parameters of interest while leveraging prior knowledge of the association between outcome variables. The methods that do allow for estimation of relationships do so typically through an error covariance matrix in multivariate regression which does not scale to other types of models. In this article we proposed the MinPEN framework to simultaneously estimate regression coefficients associated with the multivariate regression model and the relationships between outcome variables using mild assumptions. The MinPen framework utilizes a novel penalty based on the minimum function to exploit detected relationships between responses. An iterative algorithm that …

Quasiclassical Computations Of Compton-Scattered Spectra, Erik Scott Johnson

Physics Theses & Dissertations

Quality X-ray sources are crucial to fundamental physics research, medical radiology, humanities research, and materials science. While synchrotron radiation (SR) facilities produce the state-of-the- art emissions with respect to brilliance and frequency tunability, the great expense required to build, maintain, and operate these structures greatly limits their accessibility to researchers. Much of the research conducted at SR facilities, however, may be conducted with inverse Compton sources (ICS). Accelerator-based Compton scattering light sources generate high-energy, high-brilliance emissions. Compton scattering is the process by which a photon scatters off an electron. ICS offer an affordable, in-lab alternative to SR facilities. Even though …

Tess As A Low-Surface-Brightness Observatory: Cutouts From Wide-Area Coadded Images, G. Bruce Berriman, John C. Good, Benne Holwerda

Faculty and Staff Scholarship

We present a mosaic of those co-added Full Frame Images acquired by the TESS satellite that had been released in 2020 April. The mosaic shows substantial stray light over the sky. Yet over spatial scales of a few degrees, the background appears uniform. This result indicates that TESS has considerable potential as a Low Surface Brightness Observatory. The co-added images are freely available as a High Level Science Product (HLSP) at MAST and accessible through a Jupyter Notebook.

Developing An Effective Targeted Mobile Application To Enhance Transportation Safety And Use Of Active Transportation Modes In Fresno County: The Role Of Application Design & Content, Samer Sarofim

Mineta Transportation Institute

Do pedestrians and cyclists need their own app? Pedestrians and cyclists in Fresno county think so, and this research examined this need and how it relates to the importance of app design. Survey participants (all who regularly use active transportation modes) along with a variety of transportation stakeholders, including the Fresno Council of Government, the California Department of Transportation (Caltrans) District 6, and the City of Fresno — Public Works Department, indicated the importance of designing effective communication tools to enhance the utilization of active transportation modes and to ensure the safety of vulnerable road users. In this study, over …

Efficient Heuristic Solutions To Scheduling Online Courses, Rida Zaidi

Electronic Theses and Dissertations

The demand for efficient algorithms to automate (near-)optimal timetables has motivated many well-studied scheduling problems in operational research. With most of the courses moving online during the recent pandemic, the delivery of quality education has raised many new technical issues, including online course scheduling. This thesis considers the problem of yielding a near-optimal schedule of the real-time courses in an educational institute, taking into account the conflict among courses, the constraint on the simultaneous consumption of the bandwidth at the hosting servers of the courses, and the maximum utilization of the prime time for the lectures. We propose three approaches …

A Network-Based Approach For Computational Drug Repurposing On Cancer Data, Ann Reba Thomas Alexander

Electronic Theses and Dissertations

In this thesis, we are interested in finding the best drugs that can be repurposed for the disease and able to find the adverse effects such drugs that are FDA-Approved. Developing an effective drug can be a time-consuming and expensive crucible method. Network-based machine learning methods are used for predicting a given drug for A that can be used for B. It aims at finding new indications for already existing drugs and therefore increases the available therapeutic choices at a fraction of the cost of new drug development. The perturbation gene expression data corresponding to the MCF7 cell line was …

Computational Enzymology On Sulfur-Containing Enzymes: From Method To Application, Paul Meister

Electronic Theses and Dissertations

Sulfur-containing biomolecules display incredible functional diversity. Indeed, in addition to thiols and thioethers, S-nitrosothiols, 3,4-coordinate, sulfoxides, persulfides and now even polysulfides are commonly observed intermediates. Unfortunately, however, their biological synthesis and roles remain poorly understood. In addition, sulfur-containing species can access a broad range of oxidation states and thus can act as either an electrophile or nucleophile giving rise to an even more diverse set of sulfur-derived functional groups. However, these unique properties can lead to difficulties in characterizing such compounds experimentally and reinforces the need for computational studies to reliably predict their structural and energetic properties. In this dissertation, …

Improved Glove Word Embedding Using Linear Weighting Scheme For Word Similarity Tasks, Quinlan Lu

Electronic Theses and Dissertations

One of the trends in Natural Language Processing (NLP) is the use of word embedding. Its aim is to build a low dimensional vector representation of words from text corpora. Global Vectors for Word Representation (GloVe) and Sikp-Gram with Negative Sampling (SGNS) are two representative word embedding methods. Existing papers have different conclusions on the performance of these two methods. This thesis focuses on GloVe and studies its commonalities and differences with SGNS.

Word co-occurrence is the cornerstone of all word embedding algorithms. One difference between GloVe and SGNS is the definition of co-occurrence. The weight of co-occurring words tapers …

Strain Fields In Twisted Bilayer Graphene, Nathanael P. Kazmierczak, Madeline Van Winkle, Colin Ophus, Karen C. Bustillo

University Faculty Publications and Creative Works

Van der Waals heteroepitaxy allows deterministic control over lattice mismatch or azimuthal orientation between atomic layers to produce long-wavelength superlattices. The resulting electronic phases depend critically on the superlattice periodicity and localized structural deformations that introduce disorder and strain. In this study we used Bragg interferometry to capture atomic displacement fields in twisted bilayer graphene with twist angles