Mountain Scholar :: Browsing by Author "Anderson, Charles, committee member"

Browsing by Author "Anderson, Charles, committee member"

Now showing 1 - 20 of 33

Open Access
A real-time building HVAC model implemented as a tool for decision making in early stages of design
(Colorado State University. Libraries, 2015) Syed, Zaker Ali, author; Bradley, Thomas H., advisor; Anderson, Charles, committee member; Sampath, Walajabad, committee member
Construction of buildings is one of the major sources of greenhouse gases (GHGs) and energy consumption. It would therefore be beneficial to improve the design of new buildings so that they consume less energy and reduce GHG emissions over their lifecycle. However, the design of these “green buildings” is challenging because the analyses required to design and optimize these buildings is time intensive and complicated. In response, numerous software applications have been developed over the years by various government agencies, organizations and researchers. But, recent surveys of architects have found that these energy simulation programs are used irregularly and by very few architectural firms. The utility of these programs is limited by three main factors. First, these software applications are complicated, stand-alone programs that require extensive training to be effective. Second, there are a large set of energy simulation programs available, all of which have differing metrics of building performance with differing degrees of accuracy. And lastly, these applications do not fit into the conventional workflow that architects follow for a majority of projects. To address these issues, this thesis focuses on the development of a simplified HVAC model that not only gives sufficiently accurate results but also can be easily integrated into the conventional design workflow. There are some key challenges in developing such a model. • Early in the design process (when many irreversible energy impacting design decisions are made) there is very limited information available about the building materials, heat loads, and more. • The simulation must be integrated into the design software and workflows that are currently being used by architects. This requires a near-instantaneous calculation method that can extract information from the only available data at the initial design (sketching) phase, the computer aided design (CAD) models and the location. To achieve these objectives, the Radiant Time Series (RTS) method was supplemented with real data from National Solar Radiation Database to enable a near-instantaneous annual HVAC load calculation to be integrated into preliminary CAD modelling software. This model was integrated in to the Trimble Sketch-up™ software. The combined software system is demonstrated to enable effective design feedback in early design decisions including building orientation, construction material and design of fenestration.
Open Access
Accurate dimension reduction based polynomial chaos approach for uncertainty quantification of high speed networks
(Colorado State University. Libraries, 2018) Krishna Prasad, Aditi, author; Roy, Sourajeey, advisor; Pezeshki, Ali, committee member; Notaros, Branislav, committee member; Anderson, Charles, committee member
With the continued miniaturization of VLSI technology to sub-45 nm levels, uncertainty in nanoscale manufacturing processes and operating conditions have been found to translate into unpredictable system-level behavior of integrated circuits. As a result, there is a need for contemporary circuit simulation tools/solvers to model the forward propagation of device level uncertainty to the network response. Recently, techniques based on the robust generalized polynomial chaos (PC) theory have been reported for the uncertainty quantification of high-speed circuit, electromagnetic, and electronic packaging problems. The major bottleneck in all PC approaches is that the computational effort required to generate the metamodel scales in a polynomial fashion with the number of random input dimensions. In order to mitigate this poor scalability of conventional PC approaches, in this dissertation, a reduced dimensional PC approach is proposed. This PC approach is based on using a high dimensional model representation (HDMR) to quantify the relative impact of each dimension on the variance of the network response. The reduced dimensional PC approach is further extended to problems with mixed aleatory and epistemic uncertainties. In this mixed PC approach, a parameterized formulation of analysis of variance (ANOVA) is used to identify the statistically significant dimensions and subsequently perform dimension reduction. Mixed problems are however characterized by far greater number of dimensions than purely epistemic or aleatory problems, thus exacerbating the poor scalability of PC expansions. To address this issue, in this dissertation, a novel dimension fusion approach is proposed. This approach fuses the epistemic and aleatory dimensions within the same model parameter into a mixed dimension. The accuracy and efficiency of the proposed approaches are validated through multiple numerical examples.
Open Access
Anomaly detection in terrestrial hyperspectral video using variants of the RX algorithm
(Colorado State University. Libraries, 2012) Schwickerath, Anthony N., author; Kirby, Michael, advisor; Peterson, Christopher, committee member; Anderson, Charles, committee member
There is currently interest in detecting the use of chemical and biological weapons using hyperspectral sensors. Much of the research in this area assumes the spectral signature of the weapon is known in advance. Unfortunately, this may not always be the case. To obviate the reliance on a library of known target signatures, we instead view this as an anomaly detection problem. In this thesis, the RX algorithm, a benchmark anomaly detection algorithm for multi- and hyper-spectral data is reviewed, as are some standard extensions. This class of likelihood ratio test-based algorithms is generally applied to aerial imagery for the identification of man-made artifacts. As such, the model assumes that the scale is relatively consistent and that the targets (roads, cars) also have fixed sizes. We apply these methods to terrestrial video of biological and chemical aerosol plumes, where the background scale and target size both vary, and compare preliminary results. To explore the impact of parameter choice on algorithm performance, we also present an empirical study of the standard RX algorithm applied to synthetic targets of varying sizes over a range of settings.
Open Access
Application of neural networks to subseasonal to seasonal predictability in present and future climates
(Colorado State University. Libraries, 2022) Mayer, Kirsten J., author; Barnes, Elizabeth A., advisor; Hurrell, James W., committee member; Maloney, Eric D., committee member; Anderson, Charles, committee member
The Earth system is known for its lack of predictability on subseasonal to seasonal timescales (S2S; 2 weeks to a season). Yet accurate predictions on these timescales provide crucial, actionable lead times for agriculture, energy, and water management sectors. Fortunately, specific Earth system states – deemed forecasts of opportunity – can be leveraged to improve prediction skill. Our current understanding of these opportunities are rooted in our knowledge of the historical climate. Depending on societal actions, the future climate could vary drastically, and these possible futures could lead to varying changes to S2S predictability. In recent years, neural networks have been successfully applied to weather and climate prediction. With the rapid development of neural network explainability techniques, the application of neural networks now provides an opportunity to further understand our climate system as well. The research presented here demonstrates the utility of explainable neural networks for S2S prediction and predictability changes under future climates. The first study presents a novel approach for identifying forecasts of opportunity in observations using neural network confidence. It further demonstrates that neural networks can be used to gain physical insight into predictability, through neural network explainability techniques. We then employ this methodology to explore S2S predictability differences in two future scenarios: under anthropogenic climate change and stratospheric aerosol injection (SAI). In particular, we explore subseasonal predictability and forecasts of opportunity changes under anthropogenic warming compared to a historical climate in the CESM2-LE. We then investigate how future seasonal predictability may differ under SAI compared to a future without SAI deployment in the ARISE-SAI simulations. We find differences in predictability between the historical and future climates and the two future scenarios, respectively, where the largest differences in skill generally occur during forecasts of opportunity. This demonstrates that the forecast of opportunity approach, presented in the first study, is useful for identifying differences in future S2S predictability that may not have been identified if examining predictability across all predictions. Overall, these results demonstrate that neural networks are useful tools for exploring subseasonal to seasonal predictability, its sources, and future changes.
Open Access
Automatic creation of tile size selection models using neural networks
(Colorado State University. Libraries, 2010) Yuki, Tomofumi, author; Rajopadhye, Sanjay, advisor; Anderson, Charles, committee member; Casterella, Gretchen, committee member; Strout, Michelle, committee member
Tiling is a widely used loop transformation for exposing/exploiting parallelism and data locality. Effective use of tiling requires selection and tuning of the tile sizes. This is usually achieved by hand-crafting tile size selection (TSS) models that characterize the performance of the tiled program as a function of tile sizes. The best tile sizes are selected by either directly using the TSS model or by using the TSS model together with an empirical search. Hand-crafting accurate TSS models is hard, and adapting them to different architecture/compiler, or even keeping them up-to-date with respect to the evolution of a single compiler is often just as hard. Instead of hand-crafting TSS models, can we automatically learn or create them? In this paper, we show that for a specific class of programs fairly accurate TSS models can be automatically created by using a combination of simple program features, synthetic kernels, and standard machine learning techniques. The automatic TSS model generation scheme can also be directly used for adapting the model and/or keeping it up-to-date. We evaluate our scheme on six different architecture-compiler combinations (chosen from three different architectures and four different compilers). The models learned by our method have consistently shown near-optimal performance (within 5% of the optimal on average) across the tested architecture-compiler combinations.
Open Access
Case study of the real world integration of fuel cell plug-in hybrid electric vehicles and their effect on hydrogen refueling locations in the Puget Sound region
(Colorado State University. Libraries, 2014) Bucher, Jake Duvall, author; Bradley, Thomas, advisor; Anderson, Charles, committee member; Suryanarayanan, Siddarth, committee member
The personal vehicle transportation fleet relies heavily on non-renewable and pollutive sources of fuel, such as petroleum. However, with harsher restrictions from the Environmental Protection Agency's (EPA) Corporate Average Fuel Economy (CAFE) and California Air Resource Board's (CARB) Zero Emission Vehicle (ZEV) standards coupled with growing sales for alternative fueled vehicles, the automotive industry has begun to shift toward more renewable and clean sources of energy to power vehicles. The fuel cell plug-in hybrid electric vehicle (FCPHEV) architecture provides a unique and promising solution to decreasing the dependence of vehicles on petroleum and decreasing the amount of pollution emitted from tailpipes. Until recently, the FCPHEV architecture had only been developed in concept cars and paper studies. However, recent studies have confirmed the capability of the FCPHEV concept in terms of its economics, environmental benefits, and real-world viability. From this concept it becomes important to understand how daily commuters will benefit from driving a FCPHEV using real world driving data. Through the use of geographic information system (GIS) data of vehicle travel in the Puget Sound area from the National Renewable Energy Laboratory (NREL) a model of electrical and hydrogen energy consumption of a fleet of FCPHEVs can be constructed. This model can be modified to model the driving, charging and fueling habits of drivers using four different all-electric driving ranges, and using either a normal plug-in hybrid control strategy or a control strategy that focuses on highway fuel cell operation. These comparisons are used to analyze the driving habits of daily commuters while using a FCPHEV, and the effect of the FCPHEV architecture on the location of hydrogen refueling. The results of this thesis help to define FCPHEV energy management strategies and show that the FCPHEV architecture can concentrate the location of hydrogen refueling to predictable areas and aid in the development of the hydrogen refueling infrastructure.
Open Access
Categorical evidence, confidence and urgency during the integration of multi-feature information
(Colorado State University. Libraries, 2015) Braunlich, Kurt, author; Seger, Carol, advisor; Anderson, Charles, committee member; Rhodes, Matthew, committee member; Troup, Lucy, committee member
The present experiment utilized a temporally-extended categorization task to investigate the neural substrates underlying our ability to integrate information over time and across multiple stimulus features. Importantly, the design allowed differentiation of three important decision functions: 1) categorical evidence, 2) decisional confidence (the choice-independent probability that a decision will lead to a desirable state), and 3) urgency (a hypothetical signal representing a growing pressure to produce a behavioral response within each trial). In conjunction with model-based fMRI, the temporal evolution of these variables were tracked as participants deliberated about impending choices. The approach allowed investigation of the independent effects of urgency across the brain, and also the investigation of how urgency might modulate representations of categorical evidence and confidence. Representations associated with prediction errors during feedback were also investigated. Many cortical and striatal somatomotor regions tracked the dynamical evolution of categorical evidence, while many regions of the dorsal and ventral attention networks (Corbetta and Shulman, 2002) tracked decisional confidence and uncertainty. Urgency influenced activity in regions known to be associated with flexible control of the speed-accuracy trade-off (particularly the pre- SMA and striatum), and additionally modulated representations of categorical evidence and confidence. The results, therefore, link the urgency signal to two hypothetical mechanisms underling flexible control of decision thresholding (Bogacz et al., 2010): gain modulation of the striatal thresholding circuitry, and gain modulation of the integrated categorical evidence.
Open Access
Classification ensemble methods for mitigating concept drift within online data streams
(Colorado State University. Libraries, 2012) Barber, Michael J., author; Howe, Adele E., advisor; Anderson, Charles, committee member; Hoeting, Jennifer, committee member
The task of instance classification within very large data streams is challenged by both the overwhelming amount of data, and a phenomenon known as concept drift. In this research we provide a comprehensive comparison of several state of the art ensemble methods that purport to handle concept drift, and we propose two additional algorithms. Our two new methods, the AMPE and AMPE2 algorithms are then used to further our understanding of concept drift and the algorithmic factors that influence the performance of ensemble based concept drift algorithms.
Open Access
Control system design for plasma power generator
(Colorado State University. Libraries, 2022) Sankaran, Aishwarya, author; Young, Peter M., advisor; Chong, Edwin, committee member; Anderson, Charles, committee member
The purpose of this research is to develop advanced control strategies for precise control over power delivery to nonlinear plasma loads at high frequency. A high-fidelity MATLAB/Simulink simulation model was provided by Advanced Energy Industries, Inc (AE) and the data from this model was considered as the actual model under consideration. The research work requires computing a mathematical model of the plasma power generator system, analyzing and synthesizing robust controllers for individual operating points, and then developing a control system that covers the entire the grid of operating points. The modeling process involves developing computationally simple near-linear models representing relevant frequencies and operating points for the system consisting of nonlinear plasma load, RF Power Amplifier, and a Match Network. To characterize the (steady-state) mapping from power setpoint to delivered power the steady-state gains of the system are taken under consideration. Linear and nonlinear system identification procedures are used to adequately capture both the nonlinear steady-state gains and the linear dynamic model response. These near-linear or linear models with uncertainty description to characterize the robustness requirements are utilized in the second stage to develop a grid of robust controller designed at linear operating points. The controller from -synthesis design process optimizes robust performance for allowable perturbations as large as possible. It does all this while guaranteeing closed-loop stability for all allowable perturbations. The final stage of the research focuses on developing Linear Parameter Varying (LPV) controllers with non-linear offset. This single controller covers the entire operating range, including the case that the desired signals to track may vary over wide regions of the operating envelope. LPV controllers allows actual power to track the changing setpoint in a smooth manner over the entire operating range.
Open Access
Convex and non-convex optimization using centroid-encoding for visualization, classification, and feature selection
(Colorado State University. Libraries, 2022) Ghosh, Tomojit, author; Kirby, Michael, advisor; Anderson, Charles, committee member; Ben-Hur, Asa, committee member; Adams, Henry, committee member
Classification, visualization, and feature selection are the three essential tasks of machine learning. This Ph.D. dissertation presents convex and non-convex models suitable for these three tasks. We propose Centroid-Encoder (CE), an autoencoder-based supervised tool for visualizing complex and potentially large, e.g., SUSY with 5 million samples and high-dimensional datasets, e.g., GSE73072 clinical challenge data. Unlike an autoencoder, which maps a point to itself, a centroid-encoder has a modified target, i.e., the class centroid in the ambient space. We present a detailed comparative analysis of the method using various data sets and state-of-the-art techniques. We have proposed a variation of the centroid-encoder, Bottleneck Centroid-Encoder (BCE), where additional constraints are imposed at the bottleneck layer to improve generalization performance in the reduced space. We further developed a sparse optimization problem for the non-linear mapping of the centroid-encoder called Sparse Centroid-Encoder (SCE) to determine the set of discriminate features between two or more classes. The sparse model selects variables using the 1-norm applied to the input feature space. SCE extracts discriminative features from multi-modal data sets, i.e., data whose classes appear to have multiple clusters, by using several centers per class. This approach seems to have advantages over models which use a one-hot-encoding vector. We also provide a feature selection framework that first ranks each feature by its occurrence, and the optimal number of features is chosen using a validation set. CE and SCE are models based on neural network architectures and require the solution of non-convex optimization problems. Motivated by the CE algorithm, we have developed a convex optimization for the supervised dimensionality reduction technique called Centroid Component Retrieval (CCR). The CCR model optimizes a multi-objective cost by balancing two complementary terms. The first term pulls the samples of a class towards its centroid by minimizing a sample's distance from its class centroid in low dimensional space. The second term pushes the classes by maximizing the scattering volume of the ellipsoid formed by the class-centroids in embedded space. Although the design principle of CCR is similar to LDA, our experimental results show that CCR exhibits performance advantages over LDA, especially on high-dimensional data sets, e.g., Yale Faces, ORL, and COIL20. Finally, we present a linear formulation of Centroid-Encoder with orthogonality constraints, called Principal Centroid Component Analysis (PCCA). This formulation is similar to PCA, except the class labels are used to formulate the objective, resulting in the form of supervised PCA. We show the classification and visualization experiments results with this new linear tool.
Open Access
Distributed algorithms for the orchestration of stochastic discrete event simulations
(Colorado State University. Libraries, 2014) Sui, Zhiquan, author; Pallickara, Shrideep, advisor; Anderson, Charles, committee member; Böhm, Wim, committee member; Hayne, Stephen, committee member
Discrete event simulations are widely used in modeling real-world phenomena such as epidemiology, congestion analysis, weather forecasting, economic activity, and chemical reactions. The expressiveness of such simulations depends on the number and types of entities that are modeled and also the interactions that entities have with each other. In the case of stochastic simulations, these interactions are based on the concomitant probability density functions. The more exhaustively a phenomena is modeled, the greater its computational complexity and, correspondingly, the execution time. Distributed orchestration can speed-up such complex simulations. This dissertation considers the problem of distributed orchestration of stochastic discrete event simulations where the computations are irregular and the processing loads stochastic. We have designed a suite of algorithms that target alleviating imbalances between processing elements across synchronization time steps. The algorithms explore different aspects of the orchestration spectrum: static vs. dynamic, reactive vs. proactive, and deterministic vs. learning-based. The feature vector that guides our algorithms include externally observable features of the simulation such as computational footprints and hardware profiles, and features internal to the simulation such as entity states. The learning structure includes basic version of Artificial Neural Network (ANN) and an improved version of ANN. The algorithms are self-tuning and account for the state of the simulation and processing elements while coping with prediction errors. Finally, these algorithms address resource uncertainty as well. Resource uncertainty in such settings occurs due to resource failures, slowdowns, and heterogeneity. Task apportioning, speculative tasks to cope with stragglers, and checkpointing account for the quality and state of both the resource and simulation. The algorithms achieve demonstrably good performance. Despite the irregular nature of these computations, stochasticity in the processing loads, and resource uncertainty execution times are reduced by a factor of 1.8 when the number of resources is doubled.
Open Access
Diverse developmental trajectories of perineuronal nets during vertebrate nervous system construction
(Colorado State University. Libraries, 2018) Edwards, Jacob, author; Hoke, Kim, advisor; Anderson, Charles, committee member; Garrity, Deborah, committee member; Mueller, Rachel, committee member
In the central nervous system, aggregated extracellular matrix compounds known as perineuronal nets (PNNs) shape patterns of neural connectivity over development. Removing PNNs restores juvenile-like states of neural circuit plasticity and subsequent behavioral plasticity. Our current understanding of the role of PNNs in plasticity has resulted in promising therapeutic applications for many neurodegenerative diseases. To ensure safety and efficacy in such applications, we require a broad understanding of PNN function in the nervous system. The current data suggest that PNNs stabilize fundamental features of neural connectivity progressively in an ascending, or "ground-up", fashion. Stabilizing lower input processing pathways establishes a solid, reliable foundation for higher cognition. However, data on PNN development exists almost exclusively for mammals. Is, then, the ground-up model of circuit stabilization a general feature of PNNs across vertebrates? I found that developmental patterns of PNNs in fish (Poecilia reticulata), amphibians (Rhinella yunga), and reptiles (Anolis sagrei) follow diverse trajectories, often emerging first in higher forebrain processing pathways. Similarly, they associate with diverse cell populations and vary widely in structural characteristics both within and across species. While my data do not invalidate a ground-up model for mammal PNNs, they do suggest that this pattern may be an evolutionary innovation in this group, and that the broad roles of PNNs in circuit stability and neuronal physiology are complex and lineage-specific.
Open Access
Enhancing space and time efficiency of genomics in practice through sophisticated applications of the FM-Index
(Colorado State University. Libraries, 2018) Muggli, Martin D., author; McConnell, Ross, advisor; Morley, Paul S., committee member; Chitsaz, Hamid, committee member; Anderson, Charles, committee member
Genomic sequence data has become so easy to get that the computation to process it has become a bottleneck in the advancement of biological science. A data structure known as the FM-Index both compresses data and allows efficient querying, thus can be used to implement more efficient processing methods. In this work we apply advanced formulations of the FM-Index to existing problems and show our methods exceed the performance of competing tools.
Open Access
From RNA-Seq to gene annotation using the splicegrapher method
(Colorado State University. Libraries, 2013) Rogers, Mark F., author; Ben-Hur, Asa, advisor; Boucher, Christina, committee member; Anderson, Charles, committee member; Reddy, Anireddy S. N., committee member
Messenger RNA (mRNA) plays a central role in carrying out the instructions encoded in a gene. A gene's components may be combined in various ways to generate a diverse range of mRNA molecules, or transcripts, through a process called alternative splicing (AS). This allows each gene to produce different products under different conditions, such as different stages of development or in different tissues. Researchers can study the diverse set of transcripts a gene produces by sequencing its mRNA. The latest sequencing technology produces millions of short sequence reads (RNA-Seq) from mRNA transcripts, providing researchers with unprecedented opportunities to assess how genetic instructions change under different conditions. It is relatively inexpensive and easy to obtain these reads, but one limitation has been the lack of versatile methods to analyze the data. Most methods attempt to predict complete mRNA transcripts from patterns of RNA-Seq reads ascribed to a particular gene, but the short length of these reads makes transcript prediction problematic. We present a method, called SpliceGrapherXT, that takes a different approach by predicting splice graphs that capture in a single structure all the ways in which a gene's components may be assembled. Whereas other methods make predictions primarily from RNA-Seq evidence, SpliceGrapherXT uses gene annotations describing known transcripts to guide its predictions. We show that this approach allows SpliceGrapherXT to make predictions that encapsulate gene architectures more accurately than other state-of-the-art methods. This accuracy is crucial not only for updating gene annotations, but our splice graph predictions can contribute to more accurate transcript predictions as well. Finally we demonstrate that by using SpliceGrapherXT to assess AS on a genome-wide scale, we can gain new insights into the ways that specific genes and environmental conditions may impact an organism's transcriptome. SpliceGrapherXT is available for download at http://splicegrapher.sourceforge.net.
Open Access
Hypothesis-based machine learning for deep-water channel systems
(Colorado State University. Libraries, 2020) Vento, Noah Francis Ryoichi, author; Stright, Lisa, advisor; Ronayne, Michael, committee member; Anderson, Charles, committee member
Machine learning algorithms are readily being incorporated into petroleum industry workflows for use in well-log correlation, prediction of rock properties, and seismic data interpretation. However, there is a clear disconnect between sedimentology and data analytics in these workflows because sedimentologic data is largely qualitative and descriptive. Sedimentology defines stratigraphic architecture and heterogeneity, which can greatly impact reservoir quality and connectivity and thus hydrocarbon recovery. Deep-water channel systems are an example where predicting reservoir architecture is critical to mitigating risk in hydrocarbon exploration. Deep-water reservoirs are characterized by spatial and temporal variations in channel body stacking patterns, which are difficult to predict with the paucity of borehole data and low quality seismic available in these remote locations. These stacking patterns have been shown to be a key variable that controls reservoir connectivity. In this study, the gap between sedimentology and data analytics is bridged using machine learning algorithms to predict stratigraphic architecture and heterogeneity in a deep-water slope channel system. The algorithms classify variables that capture channel stacking patterns (i.e., channel positions: axis, off-axis, and margin) from a database of outcrop statistics sourced from 68 stratigraphic measured sections from outcrops of the Upper Cretaceous Tres Pasos Formation at Laguna Figueroa in the Magallanes Basin, Chile. An initial hypothesis that channel position could be predicted from 1D descriptive sedimentologic data was tested with a series of machine learning algorithms and classification schemes. The results confirmed this hypothesis as complex algorithms (i.e., random forest, XGBoost, and neural networks) achieved accuracies above 80% while less complex algorithms (i.e., decision trees) achieved lower accuracies between 60%-70%. However, certain classes were difficult for the machine learning algorithms to classify, such as the transitional off-axis class. Additionally, an interpretive classification scheme performed better (by around 10%-20% in some cases) than a geometric scheme that was devised to remove interpretation bias. However, outcrop observations reveal that the interpretive classification scheme may be an over-simplified approach and that more heterogeneity likely exists in each class as revealed by the geometric scheme. A refined hypothesis was developed that a hierarchical machine learning approach could lend deeper insight into the heterogeneity within sedimentologic classes that are difficult for an interpreter to discern by observation alone. This hierarchical analysis revealed distinct sub-classes in the margin channel position that highlight variations in margin depositional style. The conceptual impact of these varying margin styles on fluid flow and connectivity is shown.
Open Access
Implications for automation assistance in unmanned aerial system operator training
(Colorado State University. Libraries, 2012) Blitch, John G., author; Clegg, Benjamin A., advisor; Cleary, Anne, committee member; Anderson, Charles, committee member
The integration of automated modules into unmanned systems control has had a positive impact on operational effectiveness across a variety of challenging domains from battlefields and disaster areas to the National Airspace and distant planets. Despite the generally positive nature of such technological progress, however, concerns for complacency and other automation-induced detriments have been established in a growing body of empirical literature derived from both laboratory research and operational reviews. Given the military's demand for new Unmanned Aerial System (UAS) operators, there is a need to explore how such concerns might extend from the operational realm of experienced professionals into the novice training environment. An experiment was conducted to investigate the influence of automation on training efficiency using a Predator UAS simulator developed by the Air Force Research Laboratory (AFRL) in a modified replication of previous research. Participants were trained in a series of basic maneuvers, with half receiving automated support only on a subset of maneuvers. A subsequent novel landing test showed poorer performance for the group that received assistance from automation during training. Implications of these findings are discussed.
Open Access
Large margin kernel methods for calmodulin binding prediction
(Colorado State University. Libraries, 2010) Hamilton, Michael, author; Ben-Hur, Asa, advisor; Anderson, Charles, committee member; Iyer, Hariharan, committee member
Protein-protein interactions are involved in nearly all molecular processes of organisms. However direct laboratory techniques for identifying binding partners remain expensive and difficult at the proteome scale. In this work, kernel methods for predicting calmodulin binding partners and calmodulin binding sites are presented. Furthermore, we compare binary and structural support vector machines with multiple kernels defined over protein sequences.
Open Access
Large margin methods for partner specific prediction of interfaces in protein complexes
(Colorado State University. Libraries, 2014) Minhas, Fayyaz ul Amir Afsar, author; Ben-Hur, Asa, advisor; Draper, Bruce, committee member; Anderson, Charles, committee member; Snow, Christopher, committee member
The study of protein interfaces and binding sites is a very important domain of research in bioinformatics. Information about the interfaces between proteins can be used not only in understanding protein function but can also be directly employed in drug design and protein engineering. However, the experimental determination of protein interfaces is cumbersome, expensive and not possible in some cases with today's technology. As a consequence, the computational prediction of protein interfaces from sequence and structure has emerged as a very active research area. A number of machine learning based techniques have been proposed for the solution to this problem. However, the prediction accuracy of most such schemes is very low. In this dissertation we present large-margin classification approaches that have been designed to directly model different aspects of protein complex formation as well as the characteristics of available data. Most existing machine learning techniques for this task are partner-independent in nature, i.e., they ignore the fact that the binding propensity of a protein to bind to another protein is dependent upon characteristics of residues in both proteins. We have developed a pairwise support vector machine classifier called PAIRpred to predict protein interfaces in a partner-specific fashion. Due to its more detailed model of the problem, PAIRpred offers state of the art accuracy in predicting both binding sites at the protein level as well as inter-protein residue contacts at the complex level. PAIRpred uses sequence and structure conservation, local structural similarity and surface geometry, residue solvent exposure and template based features derived from the unbound structures of proteins forming a protein complex. We have investigated the impact of explicitly modeling the inter-dependencies between residues that are imposed by the overall structure of a protein during the formation of a protein complex through transductive and semi-supervised learning models. We also present a novel multiple instance learning scheme called MI-1 that explicitly models imprecision in sequence-level annotations of binding sites in proteins that bind calmodulin to achieve state of the art prediction accuracy for this task.
Open Access
Machine learning techniques for energy optimization in mobile embedded systems
(Colorado State University. Libraries, 2012) Donohoo, Brad Kyoshi, author; Pasricha, Sudeep, advisor; Anderson, Charles, committee member; Jayasumana, Anura P., committee member
Mobile smartphones and other portable battery operated embedded systems (PDAs, tablets) are pervasive computing devices that have emerged in recent years as essential instruments for communication, business, and social interactions. While performance, capabilities, and design are all important considerations when purchasing a mobile device, a long battery lifetime is one of the most desirable attributes. Battery technology and capacity has improved over the years, but it still cannot keep pace with the power consumption demands of today's mobile devices. This key limiter has led to a strong research emphasis on extending battery lifetime by minimizing energy consumption, primarily using software optimizations. This thesis presents two strategies that attempt to optimize mobile device energy consumption with negligible impact on user perception and quality of service (QoS). The first strategy proposes an application and user interaction aware middleware framework that takes advantage of user idle time between interaction events of the foreground application to optimize CPU and screen backlight energy consumption. The framework dynamically classifies mobile device applications based on their received interaction patterns, then invokes a number of different power management algorithms to adjust processor frequency and screen backlight levels accordingly. The second strategy proposes the usage of machine learning techniques to learn a user's mobile device usage pattern pertaining to spatiotemporal and device contexts, and then predict energy-optimal data and location interface configurations. By learning where and when a mobile device user uses certain power-hungry interfaces (3G, WiFi, and GPS), the techniques, which include variants of linear discriminant analysis, linear logistic regression, non-linear logistic regression, and k-nearest neighbor, are able to dynamically turn off unnecessary interfaces at runtime in order to save energy.
Open Access
Methods for network generation and spectral feature selection: especially on gene expression data
(Colorado State University. Libraries, 2019) Mankovich, Nathan, author; Kirby, Michael, advisor; Anderson, Charles, committee member; Peterson, Chris, committee member
Feature selection is an essential step in many data analysis pipelines due to its ability to remove unimportant data. We will describe how to realize a data set as a network using correlation, partial correlation, heat kernel and random edge generation methods. Then we lay out how to select features from these networks mainly leveraging the spectrum of the graph Laplacian, adjacency, and supra-adjacency matrices. We frame this work in the context of gene co-expression network analysis and proceed with a brief analysis of a small set of gene expression data for human subjects infected with the flu virus. We are able to distinguish two sets of 14-15 genes which produce two fold SSVM classification accuracies at certain times that are at least as high as classification accuracies done with more than 12,000 genes.