Matching Items (215)
Filtering by
- Genre: Academic theses
- Genre: Masters Thesis
- Creators: Davulcu, Hasan
- Member of: Theses and Dissertations
Description
In videos that contain actions performed unintentionally, agents do not achieve their desired goals. In such videos, it is challenging for computer vision systems to understand high-level concepts such as goal-directed behavior. On the other hand, from a very early age, humans are able to understand the relation between an agent and their ultimate goal even if the action gets disrupted or unintentional effects occur. Inculcating this ability in artificially intelligent agents would make them better social learners by not just learning from their own mistakes, i.e, reinforcement learning, but also learning from other's mistakes. For example, this could greatly reduce the search space for artificially intelligent agents for finding the correct action sequence when trying to achieve a new goal, since they would be able to learn from others what not to do as well as how/when actions result in undesired outcomes.To validate this ability of deep learning models to perform this task, the Weakly Augmented Oops (W-Oops) dataset is proposed, built upon the Oops dataset. W-Oops consists of 2,100 unintentional human action videos, with 44 goal-directed and 33 unintentional video-level activity labels collected through human annotations.
Inspired by previous methods on tasks such as weakly supervised action localization which show promise for achieving good localization results without ground truth segment annotations, this paper proposes a weakly supervised algorithm for localizing the goal-directed as well as the unintentional temporal region of a video using only video-level labels. In particular, an attention mechanism based strategy is employed that predicts the temporal regions which contributes the most to a classification task, leveraging solely video-level labels. Meanwhile, our designed overlap regularization allows the model to focus on distinct portions of the video for inferring the goal-directed and unintentional activity, while guaranteeing their temporal ordering. Extensive quantitative experiments verify the validity of our localization method.
ContributorsChakravarthy, Arnav (Author) / Yang, Yezhou (Thesis advisor) / Davulcu, Hasan (Committee member) / Pavlic, Theodore (Committee member) / Arizona State University (Publisher)
Created2021
Description
Tensors are commonly used for representing multi-dimensional data, such as Web graphs, sensor streams, and social networks. As a consequence of the increase in the use of tensors, tensor decomposition operations began to form the basis for many data analysis and knowledge discovery tasks, from clustering, trend detection, anomaly detection to correlationanalysis [31, 38]. It is well known that Singular Value matrix Decomposition (SVD) [9] is used to extract latent semantics for matrix data. When apply SVD to tensors, which have more than two modes, it is tensor decomposition. The two most popular tensor decomposition algorithms are the Tucker [54] and the CP [19] decompositions. Intuitively, they both generalize SVD to tensors. However, one key problem with tensor decomposition is its computational complexity which may cause system bottleneck. Therefore, two phase block-centric CP tensor decomposition (2PCP) was proposed to partition the tensor into small sub-tensors, execute sub-tensor decomposition in parallel and combine the factors from each sub-tensor into final decomposition factors through iterative rerefinement process. Consequently, I proposed Sub-tensor Impact Graph (SIG) to account for inaccuracy propagation among sub-tensors and measure the impact of decomposition of sub-tensors on the other's decomposition, Based on SIG, I proposed several optimization strategies to optimize 2PCP's phase-2 refinement process. Furthermore, I applied SIG and optimization strategies for data focus, data evolution, and focus shifting in tensor analysis. Personalized Tensor Decomposition (PTD) is proposed to account for the users focus given the observations that in many applications, the user may have a focus of interest i.e., part of the data for which the user needs high accuracy and beyond this area focus, accuracy may not be as critical. PTD takes as input one or more areas of focus and performs the decomposition in such a way that, when reconstructed, the accuracy of the tensor is boosted for these areas of focus. A related challenge of data evolution in tensor analytics is incremental tensor decomposition since re-computation of the whole tensor decomposition with each update will cause high computational costs and incur large memory overheads. Especially for applications where data evolves over time and the tensor-based analysis results need to be continuouslymaintained. To avoid re-decomposition, I propose a two-phase block-incremental CP-based tensor decomposition technique, BICP, that efficiently and effectively maintains tensor decomposition results in the presence of dynamically evolving tensor data. I further extend the research focus on user focus shift. User focus may change over time as data is evolving along the time. Although PTD is efficient, re-computation for each user preference update can be the bottleneck for the system. Therefore I propose dynamic evolving user focus tensor decomposition which can smartly reuse the existing decomposition result to improve the efficiency of evolving user focus block decomposition.
ContributorsHuang, shengyu (Author) / Candan, K. Selcuk (Thesis advisor) / Davulcu, Hasan (Committee member) / Sapino, Maria Luisa (Committee member) / Tong, Hanghang (Committee member) / Zou, Jia (Committee member) / Arizona State University (Publisher)
Created2021
Description
The proliferation of semantic data in the form of RDF (Resource Description Framework) triples demands an efficient, scalable, and distributed storage along with a highly available and fault-tolerant parallel processing strategy. There are three open issues with distributed RDF data management systems that are not well addressed altogether in existing work. First is the querying efficiency, second is that solutions are optimized for certain types of query patterns and don’t necessarily work well for all types, and third is concerned with reducing pre-processing cost. Therefore, the rapid growth of RDF data raises the need for an efficient partitioning strategy over distributed data management systems to improve SPARQL (SPARQL Protocol and RDF Query Language) query performance regardless of its pattern shape with minimized pre-processing overhead. In this context, the first contribution of this work is a distributed RDF data partitioning schema called 3CStore that extends the existing VP (Vertical Partitioning) approach by using a subset of triples from the VP tables based on different join correlations. This approach speeds up queries at the cost of additional pre-processing overhead. To solve this, a relational partitioning schema called VPExp was developed by splitting predicates based on explicit type information of objects. This approach gains a significant query performance only for the specific type of query where the object is bound to a value for a particular predicate. To get efficient query performance on a wide range of query patterns, an improved solution is proposed by extending the existing Property Table approach to Subset-Property Table and combined with the VP approach. Further investigation on distributed RDF processing and querying systems based on typical use cases led to a novel relational partitioning schema called PTP (Property Table Partitioning) that further partitions the whole Property Table into the number of unique properties to minimize query input size and join operations during query evaluation. Finally, an RDF data management system based on the SPARQL-over-SQL approach called S3QLRDF is developed that generates the optimal query execution plan using statistics of PTP tables to provide efficient SPARQL query processing on a distributed system.
ContributorsHassan, P M Mahmudul Mahmudul (Author) / Bansal, Srividya (Thesis advisor) / Bansal, Ajay (Committee member) / Davulcu, Hasan (Committee member) / Sarwat Abdelghany Aly Elsayed, Mohamed (Committee member) / Arizona State University (Publisher)
Created2021
Description
The need of effective forecasting models for multi-variate time series has been underlined by the integration of sensory technologies into essential applications such as building energy optimizations, flight monitoring, and health monitoring. To meet this requirement, time series prediction techniques have been expanded from uni-variate to multi-variate. However, due to the extended models’ poor ability to capture the intrinsic relationships among variates, naïve extensions of prediction approaches result in an unwanted rise in the cost of model learning and, more critically, a significant loss in model performance. While recurrent models like Long Short-Term Memory (LSTM) and Recurrent Neural Network Network (RNN) are designed to capture the temporal intricacies in data, their performance can soon deteriorate. First, I claim in this thesis that (a) by exploiting temporal alignments of variates to quantify the importance of the recorded variates in relation to a target variate, one can build a more accurate forecasting model. I also argue that (b) traditional time series similarity/distance functions, such as Dynamic Time Warping (DTW), which require that variates have similar absolute patterns are fundamentally ill-suited for this purpose, and that should instead quantify temporal correlation in terms of temporal alignments of key “events” impacting these series, rather than series similarity. Further, I propose that (c) while learning a temporal model with recurrence-based techniques (such as RNN and LSTM – even when leveraging attention strategies) is challenging and expensive, the better results can be obtained by coupling simpler CNNs with an adaptive variate selection strategy. Putting these together, I introduce a novel Selego framework for variate selection based on these arguments, and I experimentally evaluate the performance of the proposed approach on various forecasting models, such as LSTM, RNN, and CNN, for different top-X% percent variates and different forecasting time in the future (lead), on multiple real-world data sets. Experiments demonstrate that the proposed framework can reduce the number of recorded variates required to train predictive models by 90 - 98% while also increasing accuracy. Finally, I present a fault onset detection technique that leverages the precise baseline forecasting models trained using the Selego framework. The proposed, Selego-enabled Fault Detection Framework (FDF-Selego) has been experimentally evaluated within the context of detecting the onset of faults in the building Heating, Ventilation, and Air Conditioning (HVAC) system.
ContributorsTiwaskar, Manoj (Author) / Candan, K. Selcuk (Thesis advisor) / Sapino, Maria Luisa (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2021
Description
A new class of electronic materials from food and foodstuff was developed to form a “toolkit” for edible electronics along with inorganic materials. Electrical components like resistors, capacitors and inductors were fabricated with such materials and tested. Applicable devices such as filters, microphones and pH sensors were built with edible materials. Among the applications, a wireless edible pH sensor was optimized in terms of form factor, fabrication process and cost. This dissertation discusses the material sciences of food industry, design and fabrication of electronics and biomedical engineering by demonstrating edible electronic materials, components and devices such as filters, microphones and pH sensors. pH sensors are optimized for two different generations of design and fabrication.
ContributorsYang, Haokai (Author) / Jiang, Hanqing (Thesis advisor) / Yu, Hongbin (Thesis advisor) / Yao, Yu (Committee member) / Nian, Qiong (Committee member) / Zhuang, Houlong (Committee member) / Arizona State University (Publisher)
Created2021
Description
Extensive efforts have been devoted to understanding material failure in the last several decades. A suitable numerical method and specific failure criteria are required for failure simulation. The finite element method (FEM) is the most widely used approach for material mechanical modelling. Since FEM is based on partial differential equations, it is hard to solve problems involving spatial discontinuities, such as fracture and material interface. Due to their intrinsic characteristics of integro-differential governing equations, discontinuous approaches are more suitable for problems involving spatial discontinuities, such as lattice spring method, discrete element method, and peridynamics. A recently proposed lattice particle method is shown to have no restriction of Poisson’s ratio, which is very common in discontinuous methods. In this study, the lattice particle method is adopted to study failure problems. In addition of numerical method, failure criterion is essential for failure simulations. In this study, multiaxial fatigue failure is investigated and then applied to the adopted method. Another critical issue of failure simulation is that the simulation process is time-consuming. To reduce computational cost, the lattice particle method can be partly replaced by neural network model.First, the development of a nonlocal maximum distortion energy criterion in the framework of a Lattice Particle Model (LPM) is presented for modeling of elastoplastic materials. The basic idea is to decompose the energy of a discrete material point into dilatational and distortional components, and plastic yielding of bonds associated with this material point is assumed to occur only when the distortional component reaches a critical value. Then, two multiaxial fatigue models are proposed for random loading and biaxial tension-tension loading, respectively. Following this, fatigue cracking in homogeneous and composite materials is studied using the lattice particle method and the proposed multiaxial fatigue model. Bi-phase material fatigue crack simulation is performed. Next, an integration of an efficient deep learning model and the lattice particle method is presented to predict fracture pattern for arbitrary microstructure and loading conditions. With this integration, computational accuracy and efficiency are both considered. Finally, some conclusion and discussion based on this study are drawn.
ContributorsWei, Haoyang (Author) / Liu, Yongming (Thesis advisor) / Chattopadhyay, Aditi (Committee member) / Jiang, Hanqing (Committee member) / Jiao, Yang (Committee member) / Oswald, Jay (Committee member) / Arizona State University (Publisher)
Created2021
Description
Special thermal interface materials are required for connecting devices that operate at high temperatures up to 300°C. Because devices used in power electronics, such as GaN, SiC, and other wide bandgap semiconductors, can reach very high temperatures (beyond 250°C), a high melting point, and high thermal & electrical conductivity are required for the thermal interface material. Traditional solder materials for packaging cannot be used for these applications as they do not meet these requirements. Sintered nano-silver is a good candidate on account of its high thermal and electrical conductivity and very high melting point. The high temperature operating conditions of these devices lead to very high thermomechanical stresses that can adversely affect performance and also lead to failure. A number of these devices are mission critical and, therefore, there is a need for very high reliability. Thus, computational and nondestructive techniques and design methodology are needed to determine, characterize, and design the packages. Actual thermal cycling tests can be very expensive and time consuming. It is difficult to build test vehicles in the lab that are very close to the production level quality and therefore making comparisons or making predictions becomes a very difficult exercise. Virtual testing using a Finite Element Analysis (FEA) technique can serve as a good alternative. In this project, finite element analysis is carried out to help achieve this objective. A baseline linear FEA is performed to determine the nature and magnitude of stresses and strains that occur during the sintering step. A nonlinear coupled thermal and mechanical analysis is conducted for the sintering step to study the behavior more accurately and in greater detail. Damage and fatigue analysis are carried out for multiple thermal cycling conditions. The results are compared with the actual results from a prior study. A process flow chart outlining the FEA modeling process is developed as a template for the future work. A Coffin-Manson type relationship is developed to help determine the accelerated aging conditions and predict life for different service conditions.
ContributorsAmla, Tarun (Author) / Chawla, Nikhilesh (Thesis advisor) / Jiao, Yang (Committee member) / Liu, Yongming (Committee member) / Zhuang, Houlong (Committee member) / Jiang, Hanqing (Committee member) / Arizona State University (Publisher)
Created2020
Description
Many real-world problems, such as model- and data-driven computer simulation analysis, social and collaborative network analysis, brain data analysis, and so on, benefit from jointly modeling and analyzing the underlying patterns associated with complex, multi-relational data. Tensor decomposition is an ideal mathematical tool for this joint modeling, due to its simultaneous analysis of such multi-relational data, which is made possible by the data's multidimensional, array-based nature. A major challenge in tensor decomposition lies with its computational and space complexity, especially for dense datasets. While the process is comparatively faster for sparse tensors, decomposition is still a major bottleneck for many applications. The tensor decomposition process results in dense (hence, large) intermediate results, even when the input tensor is sparse (or small). Noise is another challenge for most data mining techniques, and many tensor decomposition schemes are sensitive to noisy datasets; this is an inevitable problem for real-world data, which can lead to false conclusions. In this dissertation, I develop innovative tensor decomposition algorithms for mining both sparse and dense multi-relational data in a noise-resistant way. I present novel, scalable, parallelizable tensor decomposition algorithms, specifically tuned to be effective for dense, noisy tensors, and which maintain the quality of the resulting analysis. Furthermore, I present results on multi-relational data applications focusing on model- and data-driven computer simulation analysis, as well as social network and web mining, which demonstrate the effectiveness of these tensor decompositions.
ContributorsLi, Xinsheng (Author) / Candan, Kasim S (Thesis advisor) / Davulcu, Hasan (Committee member) / Sapino, Maria L (Committee member) / Tong, Hanghang (Committee member) / Arizona State University (Publisher)
Created2019
Description
Emerging trends in cyber system security breaches in critical cloud infrastructures show that attackers have abundant resources (human and computing power), expertise and support of large organizations and possible foreign governments. In order to greatly improve the protection of critical cloud infrastructures, incorporation of human behavior is needed to predict potential security breaches in critical cloud infrastructures. To achieve such prediction, it is envisioned to develop a probabilistic modeling approach with the capability of accurately capturing system-wide causal relationship among the observed operational behaviors in the critical cloud infrastructure and accurately capturing probabilistic human (users’) behaviors on subsystems as the subsystems are directly interacting with humans. In our conceptual approach, the system-wide causal relationship can be captured by the Bayesian network, and the probabilistic human behavior in the subsystems can be captured by the Markov Decision Processes. The interactions between the dynamically changing state graphs of Markov Decision Processes and the dynamic causal relationships in Bayesian network are key components in such probabilistic modelling applications. In this thesis, two techniques are presented for supporting the above vision to prediction of potential security breaches in critical cloud infrastructures. The first technique is for evaluation of the conformance of the Bayesian network with the multiple MDPs. The second technique is to evaluate the dynamically changing Bayesian network structure for conformance with the rules of the Bayesian network using a graph checker algorithm. A case study and its simulation are presented to show how the two techniques support the specific parts in our conceptual approach to predicting system-wide security breaches in critical cloud infrastructures.
ContributorsNagaraja, Vinjith (Author) / Yau, Stephen S. (Thesis advisor) / Ahn, Gail-Joon (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2015
Description
The purpose of this research is to efficiently analyze certain data provided and to see if a useful trend can be observed as a result. This trend can be used to analyze certain probabilities. There are three main pieces of data which are being analyzed in this research: The value for δ of the call and put option, the %B value of the stock, and the amount of time until expiration of the stock option. The %B value is the most important. The purpose of analyzing the data is to see the relationship between the variables and, given certain values, what is the probability the trade makes money. This result will be used in finding the probability certain trades make money over a period of time.
Since options are so dependent on probability, this research specifically analyzes stock options rather than stocks themselves. Stock options have value like stocks except options are leveraged. The most common model used to calculate the value of an option is the Black-Scholes Model [1]. There are five main variables the Black-Scholes Model uses to calculate the overall value of an option. These variables are θ, δ, γ, v, and ρ. The variable, θ is the rate of change in price of the option due to time decay, δ is the rate of change of the option’s price due to the stock’s changing value, γ is the rate of change of δ, v represents the rate of change of the value of the option in relation to the stock’s volatility, and ρ represents the rate of change in value of the option in relation to the interest rate [2]. In this research, the %B value of the stock is analyzed along with the time until expiration of the option. All options have the same δ. This is due to the fact that all the options analyzed in this experiment are less than two months from expiration and the value of δ reveals how far in or out of the money an option is.
The machine learning technique used to analyze the data and the probability
is support vector machines. Support vector machines analyze data that can be classified in one of two or more groups and attempts to find a pattern in the data to develop a model, which reliably classifies similar, future data into the correct group. This is used to analyze the outcome of stock options.
ContributorsReeves, Michael (Author) / Richa, Andrea (Thesis advisor) / McCarville, Daniel R. (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2015