Matching Items (37)
Description
The rapid escalation of technology and the widespread emergence of modern technological equipments have resulted in the generation of humongous amounts of digital data (in the form of images, videos and text). This has expanded the possibility of solving real world problems using computational learning frameworks. However, while gathering a large amount of data is cheap and easy, annotating them with class labels is an expensive process in terms of time, labor and human expertise. This has paved the way for research in the field of active learning. Such algorithms automatically select the salient and exemplar instances from large quantities of unlabeled data and are effective in reducing human labeling effort in inducing classification models. To utilize the possible presence of multiple labeling agents, there have been attempts towards a batch mode form of active learning, where a batch of data instances is selected simultaneously for manual annotation. This dissertation is aimed at the development of novel batch mode active learning algorithms to reduce manual effort in training classification models in real world multimedia pattern recognition applications. Four major contributions are proposed in this work: $(i)$ a framework for dynamic batch mode active learning, where the batch size and the specific data instances to be queried are selected adaptively through a single formulation, based on the complexity of the data stream in question, $(ii)$ a batch mode active learning strategy for fuzzy label classification problems, where there is an inherent imprecision and vagueness in the class label definitions, $(iii)$ batch mode active learning algorithms based on convex relaxations of an NP-hard integer quadratic programming (IQP) problem, with guaranteed bounds on the solution quality and $(iv)$ an active matrix completion algorithm and its application to solve several variants of the active learning problem (transductive active learning, multi-label active learning, active feature acquisition and active learning for regression). These contributions are validated on the face recognition and facial expression recognition problems (which are commonly encountered in real world applications like robotics, security and assistive technology for the blind and the visually impaired) and also on collaborative filtering applications like movie recommendation.
ContributorsChakraborty, Shayok (Author) / Panchanathan, Sethuraman (Thesis advisor) / Balasubramanian, Vineeth N. (Committee member) / Li, Baoxin (Committee member) / Mittelmann, Hans (Committee member) / Ye, Jieping (Committee member) / Arizona State University (Publisher)
Created2013
Description
Traditional approaches to modeling microgrids include the behavior of each inverter operating in a particular network configuration and at a particular operating point. Such models quickly become computationally intensive for large systems. Similarly, traditional approaches to control do not use advanced methodologies and suffer from poor performance and limited operating range. In this document a linear model is derived for an inverter connected to the Thevenin equivalent of a microgrid. This model is then compared to a nonlinear simulation model and analyzed using the open and closed loop systems in both the time and frequency domains. The modeling error is quantified with emphasis on its use for controller design purposes. Control design examples are given using a Glover McFarlane controller, gain sched- uled Glover McFarlane controller, and bumpless transfer controller which are compared to the standard droop control approach. These examples serve as a guide to illustrate the use of multi-variable modeling techniques in the context of robust controller design and show that gain scheduled MIMO control techniques can extend the operating range of a microgrid. A hardware implementation is used to compare constant gain droop controllers with Glover McFarlane controllers and shows a clear advantage of the Glover McFarlane approach.
ContributorsSteenis, Joel (Author) / Ayyanar, Raja (Thesis advisor) / Mittelmann, Hans (Committee member) / Tsakalis, Konstantinos (Committee member) / Tylavsky, Daniel (Committee member) / Arizona State University (Publisher)
Created2013
Description
In many fields one needs to build predictive models for a set of related machine learning tasks, such as information retrieval, computer vision and biomedical informatics. Traditionally these tasks are treated independently and the inference is done separately for each task, which ignores important connections among the tasks. Multi-task learning aims at simultaneously building models for all tasks in order to improve the generalization performance, leveraging inherent relatedness of these tasks. In this thesis, I firstly propose a clustered multi-task learning (CMTL) formulation, which simultaneously learns task models and performs task clustering. I provide theoretical analysis to establish the equivalence between the CMTL formulation and the alternating structure optimization, which learns a shared low-dimensional hypothesis space for different tasks. Then I present two real-world biomedical informatics applications which can benefit from multi-task learning. In the first application, I study the disease progression problem and present multi-task learning formulations for disease progression. In the formulations, the prediction at each point is a regression task and multiple tasks at different time points are learned simultaneously, leveraging the temporal smoothness among the tasks. The proposed formulations have been tested extensively on predicting the progression of the Alzheimer's disease, and experimental results demonstrate the effectiveness of the proposed models. In the second application, I present a novel data-driven framework for densifying the electronic medical records (EMR) to overcome the sparsity problem in predictive modeling using EMR. The densification of each patient is a learning task, and the proposed algorithm simultaneously densify all patients. As such, the densification of one patient leverages useful information from other patients.
ContributorsZhou, Jiayu (Author) / Ye, Jieping (Thesis advisor) / Mittelmann, Hans (Committee member) / Li, Baoxin (Committee member) / Wang, Yalin (Committee member) / Arizona State University (Publisher)
Created2014
Description
This thesis considers two problems in the control of robotic swarms. Firstly, it addresses a trajectory planning and task allocation problem for a swarm of resource-constrained robots that cannot localize or communicate with each other and that exhibit stochasticity in their motion and task switching policies. We model the population dynamics of the robotic swarm as a set of advection-diffusion- reaction (ADR) partial differential equations (PDEs).
Specifically, we consider a linear parabolic PDE model that is bilinear in the robots' velocity and task-switching rates. These parameters constitute a set of time-dependent control variables that can be optimized and transmitted to the robots prior to their deployment or broadcasted in real time. The planning and allocation problem can then be formulated as a PDE-constrained optimization problem, which we solve using techniques from optimal control. Simulations of a commercial pollination scenario validate the ability of our control approach to drive a robotic swarm to achieve predefined spatial distributions of activity over a closed domain, which may contain obstacles. Secondly, we consider a mapping problem wherein a robotic swarm is deployed over a closed domain and it is necessary to reconstruct the unknown spatial distribution of a feature of interest. The ADR-based primitives result in a coefficient identification problem for the corresponding system of PDEs. To deal with the inherent ill-posedness of the problem, we frame it as an optimization problem. We validate our approach through simulations and show that reconstruction of the spatially-dependent coefficient can be achieved with considerable accuracy using temporal information alone.
Specifically, we consider a linear parabolic PDE model that is bilinear in the robots' velocity and task-switching rates. These parameters constitute a set of time-dependent control variables that can be optimized and transmitted to the robots prior to their deployment or broadcasted in real time. The planning and allocation problem can then be formulated as a PDE-constrained optimization problem, which we solve using techniques from optimal control. Simulations of a commercial pollination scenario validate the ability of our control approach to drive a robotic swarm to achieve predefined spatial distributions of activity over a closed domain, which may contain obstacles. Secondly, we consider a mapping problem wherein a robotic swarm is deployed over a closed domain and it is necessary to reconstruct the unknown spatial distribution of a feature of interest. The ADR-based primitives result in a coefficient identification problem for the corresponding system of PDEs. To deal with the inherent ill-posedness of the problem, we frame it as an optimization problem. We validate our approach through simulations and show that reconstruction of the spatially-dependent coefficient can be achieved with considerable accuracy using temporal information alone.
ContributorsElamvazhuthi, Karthik (Author) / Berman, Spring Melody (Thesis advisor) / Peet, Matthew Monnig (Committee member) / Mittelmann, Hans (Committee member) / Arizona State University (Publisher)
Created2014
Description
A new method of adaptive mesh generation for the computation of fluid flows is investigated. The method utilizes gradients of the flow solution to adapt the size and stretching of elements or volumes in the computational mesh as is commonly done in the conventional Hessian approach. However, in the new method, higher-order gradients are used in place of the Hessian. The method is applied to the finite element solution of the incompressible Navier-Stokes equations on model problems. Results indicate that a significant efficiency benefit is realized.
ContributorsShortridge, Randall (Author) / Chen, Kang Ping (Thesis advisor) / Herrmann, Marcus (Thesis advisor) / Wells, Valana (Committee member) / Huang, Huei-Ping (Committee member) / Mittelmann, Hans (Committee member) / Arizona State University (Publisher)
Created2011
Description
Project portfolio selection (PPS) is a significant problem faced by most organizations. How to best select the many innovative ideas that a company has developed to deploy in a proper and sustained manner with a balanced allocation of its resources over multiple time periods is one of vital importance to a company's goals. This dissertation details the steps involved in deploying a more intuitive portfolio selection framework that facilitates bringing analysts and management to a consensus on ongoing company efforts and buy into final decisions. A binary integer programming selection model that constructs an efficient frontier allows the evaluation of portfolios on many different criteria and allows decision makers (DM) to bring their experience and insight to the table when making a decision is discussed. A binary fractional integer program provides additional choices by optimizing portfolios on cost-benefit ratios over multiple time periods is also presented. By combining this framework with an `elimination by aspects' model of decision making, DMs evaluate portfolios on various objectives and ensure the selection of a portfolio most in line with their goals. By presenting a modeling framework to easily model a large number of project inter-dependencies and an evolutionary algorithm that is intelligently guided in the search for attractive portfolios by a beam search heuristic, practitioners are given a ready recipe to solve big problem instances to generate attractive project portfolios for their organizations. Finally, this dissertation attempts to address the problem of risk and uncertainty in project portfolio selection. After exploring the selection of portfolios based on trade-offs between a primary benefit and a primary cost, the third important dimension of uncertainty of outcome and the risk a decision maker is willing to take on in their quest to select the best portfolio for their organization is examined.
ContributorsSampath, Siddhartha (Author) / Gel, Esma (Thesis advisor) / Fowler, Jown W (Thesis advisor) / Kempf, Karl G. (Committee member) / Pan, Rong (Committee member) / Sefair, Jorge (Committee member) / Arizona State University (Publisher)
Created2018
Description
This research develops heuristics to manage both mandatory and optional network capacity reductions to better serve the network flows. The main application discussed relates to transportation networks, and flow cost relates to travel cost of users of the network. Temporary mandatory capacity reductions are required by maintenance activities. The objective of managing maintenance activities and the attendant temporary network capacity reductions is to schedule the required segment closures so that all maintenance work can be completed on time, and the total flow cost over the maintenance period is minimized for different types of flows. The goal of optional network capacity reduction is to selectively reduce the capacity of some links to improve the overall efficiency of user-optimized flows, where each traveler takes the route that minimizes the traveler’s trip cost. In this dissertation, both managing mandatory and optional network capacity reductions are addressed with the consideration of network-wide flow diversions due to changed link capacities.
This research first investigates the maintenance scheduling in transportation networks with service vehicles (e.g., truck fleets and passenger transport fleets), where these vehicles are assumed to take the system-optimized routes that minimize the total travel cost of the fleet. This problem is solved with the randomized fixed-and-optimize heuristic developed. This research also investigates the maintenance scheduling in networks with multi-modal traffic that consists of (1) regular human-driven cars with user-optimized routing and (2) self-driving vehicles with system-optimized routing. An iterative mixed flow assignment algorithm is developed to obtain the multi-modal traffic assignment resulting from a maintenance schedule. The genetic algorithm with multi-point crossover is applied to obtain a good schedule.
Based on the Braess’ paradox that removing some links may alleviate the congestion of user-optimized flows, this research generalizes the Braess’ paradox to reduce the capacity of selected links to improve the efficiency of the resultant user-optimized flows. A heuristic is developed to identify links to reduce capacity, and the corresponding capacity reduction amounts, to get more efficient total flows. Experiments on real networks demonstrate the generalized Braess’ paradox exists in reality, and the heuristic developed solves real-world test cases even when commercial solvers fail.
This research first investigates the maintenance scheduling in transportation networks with service vehicles (e.g., truck fleets and passenger transport fleets), where these vehicles are assumed to take the system-optimized routes that minimize the total travel cost of the fleet. This problem is solved with the randomized fixed-and-optimize heuristic developed. This research also investigates the maintenance scheduling in networks with multi-modal traffic that consists of (1) regular human-driven cars with user-optimized routing and (2) self-driving vehicles with system-optimized routing. An iterative mixed flow assignment algorithm is developed to obtain the multi-modal traffic assignment resulting from a maintenance schedule. The genetic algorithm with multi-point crossover is applied to obtain a good schedule.
Based on the Braess’ paradox that removing some links may alleviate the congestion of user-optimized flows, this research generalizes the Braess’ paradox to reduce the capacity of selected links to improve the efficiency of the resultant user-optimized flows. A heuristic is developed to identify links to reduce capacity, and the corresponding capacity reduction amounts, to get more efficient total flows. Experiments on real networks demonstrate the generalized Braess’ paradox exists in reality, and the heuristic developed solves real-world test cases even when commercial solvers fail.
ContributorsPeng, Dening (Author) / Mirchandani, Pitu B. (Thesis advisor) / Sefair, Jorge (Committee member) / Wu, Teresa (Committee member) / Zhou, Xuesong (Committee member) / Arizona State University (Publisher)
Created2017
Description
The Kuramoto model is an archetypal model for studying synchronization in groups
of nonidentical oscillators where oscillators are imbued with their own frequency and
coupled with other oscillators though a network of interactions. As the coupling
strength increases, there is a bifurcation to complete synchronization where all oscillators
move with the same frequency and show a collective rhythm. Kuramoto-like
dynamics are considered a relevant model for instabilities of the AC-power grid which
operates in synchrony under standard conditions but exhibits, in a state of failure,
segmentation of the grid into desynchronized clusters.
In this dissertation the minimum coupling strength required to ensure total frequency
synchronization in a Kuramoto system, called the critical coupling, is investigated.
For coupling strength below the critical coupling, clusters of oscillators form
where oscillators within a cluster are on average oscillating with the same long-term
frequency. A unified order parameter based approach is developed to create approximations
of the critical coupling. Some of the new approximations provide strict lower
bounds for the critical coupling. In addition, these approximations allow for predictions
of the partially synchronized clusters that emerge in the bifurcation from the
synchronized state.
Merging the order parameter approach with graph theoretical concepts leads to a
characterization of this bifurcation as a weighted graph partitioning problem on an
arbitrary networks which then leads to an optimization problem that can efficiently
estimate the partially synchronized clusters. Numerical experiments on random Kuramoto
systems show the high accuracy of these methods. An interpretation of the
methods in the context of power systems is provided.
of nonidentical oscillators where oscillators are imbued with their own frequency and
coupled with other oscillators though a network of interactions. As the coupling
strength increases, there is a bifurcation to complete synchronization where all oscillators
move with the same frequency and show a collective rhythm. Kuramoto-like
dynamics are considered a relevant model for instabilities of the AC-power grid which
operates in synchrony under standard conditions but exhibits, in a state of failure,
segmentation of the grid into desynchronized clusters.
In this dissertation the minimum coupling strength required to ensure total frequency
synchronization in a Kuramoto system, called the critical coupling, is investigated.
For coupling strength below the critical coupling, clusters of oscillators form
where oscillators within a cluster are on average oscillating with the same long-term
frequency. A unified order parameter based approach is developed to create approximations
of the critical coupling. Some of the new approximations provide strict lower
bounds for the critical coupling. In addition, these approximations allow for predictions
of the partially synchronized clusters that emerge in the bifurcation from the
synchronized state.
Merging the order parameter approach with graph theoretical concepts leads to a
characterization of this bifurcation as a weighted graph partitioning problem on an
arbitrary networks which then leads to an optimization problem that can efficiently
estimate the partially synchronized clusters. Numerical experiments on random Kuramoto
systems show the high accuracy of these methods. An interpretation of the
methods in the context of power systems is provided.
ContributorsGilg, Brady (Author) / Armbruster, Dieter (Thesis advisor) / Mittelmann, Hans (Committee member) / Scaglione, Anna (Committee member) / Strogatz, Steven (Committee member) / Welfert, Bruno (Committee member) / Arizona State University (Publisher)
Created2018
Description
Researchers and practitioners have widely studied road network traffic data in different areas such as urban planning, traffic prediction and spatial-temporal databases. For instance, researchers use such data to evaluate the impact of road network changes. Unfortunately, collecting large-scale high-quality urban traffic data requires tremendous efforts because participating vehicles must install Global Positioning System(GPS) receivers and administrators must continuously monitor these devices. There have been some urban traffic simulators trying to generate such data with different features. However, they suffer from two critical issues (1) Scalability: most of them only offer single-machine solution which is not adequate to produce large-scale data. Some simulators can generate traffic in parallel but do not well balance the load among machines in a cluster. (2) Granularity: many simulators do not consider microscopic traffic situations including traffic lights, lane changing, car following. This paper proposed GeoSparkSim, a scalable traffic simulator which extends Apache Spark to generate large-scale road network traffic datasets with microscopic traffic simulation. The proposed system seamlessly integrates with a Spark-based spatial data management system, GeoSpark, to deliver a holistic approach that allows data scientists to simulate, analyze and visualize large-scale urban traffic data. To implement microscopic traffic models, GeoSparkSim employs a simulation-aware vehicle partitioning method to partition vehicles among different machines such that each machine has a balanced workload. The experimental analysis shows that GeoSparkSim can simulate the movements of 200 thousand cars over an extensive road network (250 thousand road junctions and 300 thousand road segments).
ContributorsFu, Zishan (Author) / Sarwat, Mohamed (Thesis advisor) / Pedrielli, Giulia (Committee member) / Sefair, Jorge (Committee member) / Arizona State University (Publisher)
Created2019
Description
The shift in focus of manufacturing systems to high-mix and low-volume production poses a challenge to both efficient scheduling of manufacturing operations and effective assessment of production capacity. This thesis considers the problem of scheduling a set of jobs that require machine and worker resources to complete their manufacturing operations. Although planners in manufacturing contexts typically focus solely on machines, schedules that only consider machining requirements may be problematic during implementation because machines need skilled workers and cannot run unsupervised. The model used in this research will be beneficial to these environments as planners would be able to determine more realistic assignments and operation sequences to minimize the total time required to complete all jobs. This thesis presents a mathematical formulation for concurrent scheduling of machines and workers that can optimally schedule a set of jobs while accounting for changeover times between operations. The mathematical formulation is based on disjunctive constraints that capture the conflict between operations when trying to schedule them to be performed by the same machine or worker. An additional formulation extends the previous one to consider how cross-training may impact the production capacity and, for a given budget, provide training recommendations for specific workers and operations to reduce the makespan. If training a worker is advantageous to increase production capacity, the model recommends the best time window to complete it such that overlaps with work assignments are avoided. It is assumed that workers can perform tasks involving the recently acquired skills as soon as training is complete. As an alternative to the mixed-integer programming formulations, this thesis provides a math-heuristic approach that fixes the order of some operations based on Largest Processing Time (LPT) and Shortest Processing Time (SPT) procedures, while allowing the exact formulation to find the optimal schedule for the remaining operations. Computational experiments include the use of the solution for the no-training problem as a starting feasible solution to the training problem. Although the models provided are general, the manufacturing of Printed Circuit Boards are used as a case study.
ContributorsAdams, Katherine Bahia (Author) / Sefair, Jorge (Thesis advisor) / Askin, Ronald (Thesis advisor) / Webster, Scott (Committee member) / Arizona State University (Publisher)
Created2019