No description available.

Published Article / A component, which has a perfect combination of different materials (probably including homogeneous materials and three different types of heterogeneous materials) in its different portions for a specific application, is considered as the component made of a multiphase perfect material. To fabricate such components, a hybrid layered manufacturing process has been developed. In order to accurately spray different materials with their required volume fractions for every pixel during fabrication, it is important to investigate its spraying operation. This paper establishes the behavior model of the spraying device and proves its validity using digital simulations.

Most computer game characters are either ``talkers,' i.e., they engage in dialogue with the player, or ``fighters,' i.e., they engage in actions against or with the player, and that may affect the virtual world. The reason for this dichotomy is a corresponding gap in the underlying development technologies used for each kind of character. Using concepts from task modeling and computational linguistics, we have developed a new kind of character-authoring technology which bridges this gap, thereby making it possible to create richer and more interesting characters for computer games.

Graduate teaching assistants (GTAs) face the unknown as they negotiate their multiple roles and identities within the graduate school and classroom setting as teachers, students, and researchers. The purpose of this study is to identify the role that institutionalized socialization, social support, and behavioral observation and modeling play for GTAs as they navigate their way through the organizational socialization process. Interviews with twenty two current and former graduate teaching assistants from a Communication department at a large, southeastern university (GSU) were conducted and analyzed. Findings indicate that institutionalized socialization, which exists at both the graduate school and departmental level, serves to both reduce and create uncertainty and anxiety for GTAs based on messages communicated and also serves the purpose of relationship formation. In examining the social support aspect, findings indicate that the socialization process is facilitated for GTAs through House‘s (1981) four categories of emotional, instrumental, informational, and appraisal support. Finally, behavioral observation aids in the socialization process for GTAs. Observation is used by GTAs to obtain information about teaching behaviors, specifically what they should and should not do in the GSU classroom. Observation also highlighted both positive and negative aspects of the departmental culture and helped GTAs to understand how things work in the department. Implications, limitations, ideas for what can be done to improve the process for GTAs, and areas for future research are also discussed.

Failure is a necessary step in the process of learning. For this reason, there has been a myriad of research dedicated to the study of student perseverance in the presence of failure, leading to several commonly-cited theories and frameworks to characterize productive and unproductive representations of the construct of persistence. While researchers are in agreement that it is important for students to persist when struggling to learn new material, there can be both positive and negative aspects of persistence. What is it, then, that separates productive from unproductive persistence? The purpose of this work is to address this question through the development, extension, and study of data-driven models of student affect, behavior, and knowledge. The increased adoption of computer-based learning platforms in real classrooms has led to unique opportunities to study student learning at both fine levels of granularity and longitudinally at scale. Prior work has leveraged machine learning methods, existing learning theory, and previous education research to explore various aspects of student learning. These include the development of sensor-free detectors that utilize only the student interaction data collected through such learning platforms. Building off of the considerable amount of prior research, this work employs state-of-the-art machine learning methods in conjunction with the large scale granular data collected by computer-based learning platforms in alignment with three goals. First, this work focuses on the development of student models that study learning through the use of advancements in student modeling and deep learning methodologies. Second, this dissertation explores the development of tools that incorporate such models to support teachers in taking action in real classrooms to promote productive approaches to learning. Finally, this work aims to complete the loop in utilizing these detector models to better understand the underlying constructs that are being measured through their application and their connection to productive perseverance and commonly-observed learning outcomes.

Human behavior plays an important role in infectious disease epidemics. The choice of preventive actions taken by individuals can completely change the epidemic outcome. Computational epidemiologists usually employ large-scale agent-based simulations of human populations to study disease outbreaks and assess intervention strategies. Such simulations rarely take into account the decision-making process of human beings when it comes to preventive behaviors. Absence of realistic agent behavior can undermine the reliability of insights generated by such simulations and might make them ill-suited for informing public health policies. In this thesis, we address this problem by developing a methodology to create and calibrate an agent decision-making model for a large multi-agent simulation, in a data driven way. Our method optimizes a cost vector associated with the various behaviors to match the behavior distributions observed in a detailed survey of human behaviors during influenza outbreaks. Our approach is a data-driven way of incorporating decision making for agents in large-scale epidemic simulations. / Master of Science / In the real world, individuals can decide to adopt certain behaviors that reduce their chances of contracting a disease. For example, using hand sanitizers can reduce an individual‘s chances of getting infected by influenza. These behavioral decisions, when taken by many individuals in the population, can completely change the course of the disease. Such behavioral decision-making is generally not considered during in-silico simulations of infectious diseases. In this thesis, we address this problem by developing a methodology to create and calibrate a decision making model that can be used by agents (i.e., synthetic representations of humans in simulations) in a data driven way. Our method also finds a cost associated with such behaviors and matches the distribution of behavior observed in the real world with that observed in a survey. Our approach is a data-driven way of incorporating decision making for agents in large-scale epidemic simulations.

The need to develop appropriate computer training techniques has led to an explosion of research in information systems. One of the most recent studies was conducted by Compeau and Higgins (1995) in which two training methods were examined in the context of Social Cognitive Learning Theory (Bandura, 1977, 1978). The training methods examined were traditional lecture-based training and behavior modeling. Based on various anomalies found in the Compeau and Higgins study, this paper introduced a moderating variable, task complexity, into their model and also attempted to replicate their original experiment. This study also incorporated an additional training method, Computer Aided Instruction (CAI), which was examined in an experiment by Gist, Schwoerer, and Rosen (1989). It was hypothesized that task complexity has a moderating effect on the relationships between behavior modeling and performance, between behavior modeling and self-efficacy, and between self-efficacy and performance. Finally, an empirical investigation was performed to determine the relative effectiveness of the three training methods examined. To test these hypothesized relationships, an experiment was conducted that examined prior performance, self-efficacy, outcome expectations, and actual performance at two levels of task complexity for each of the three training methods. The data were analyzed using a combination of multivariate and univariate analyses of variance and a structural equation modeling software package, AMOS©. Five of the original nine hypotheses from the Compeau and Higgins study were fully supported; however, none of the task complexity and only one of the avoidance behavior hypotheses were supported. Possible causes of this lack of support were multi-dimensionality of constructs or the need to examine task dimensions other than complexity. Relevant findings in this study included (1) a positive significant relationship between behavior modeling and final performance, (2) a positive significant relationship between prior performance and the endogenous constructs in the model: computer self-efficacy, outcome expectations, and final performance, and (3) a ranking of the three training methods in terms of effectiveness. Although behavior modeling produced the best performance results at all levels of task complexity, CAI was equally effective when the level of complexity was high. For low complexity tasks, however, CAI was the least effective method examined. / Ph. D.

Motivated by challenges and opportunities in nutritional epidemiology and food journaling, ubiquitous computing researchers have proposed numerous techniques for automated dietary monitoring (ADM) over the years. Although progress has been made, a truly practical system that can automatically recognize what people eat in real-world settings remains elusive. This dissertation addresses the problem of ADM by focusing on practical eating moment detection. Eating detection is a foundational element of ADM since automatically recognizing when a person is eating is required before identifying what and how much is being consumed. Additionally, eating detection can serve as the basis for new types of dietary self-monitoring practices such as semi-automated food journaling. In this thesis, I show that everyday eating moments such as breakfast, lunch, and dinner can be automatically detected in real-world settings by opportunistically leveraging sensors in practical, off-the-shelf wearable devices. I refer to this instrumentation approach as "commodity sensing". The work covered by this thesis encompasses a series of experiments I conducted with a total of 106 participants where I explored a variety of sensing modalities for automatic eating moment detection. The modalities studied include first-person images taken with wearable cameras, ambient sounds, and on-body inertial sensors. I discuss the extent to which first-person images reflecting everyday experiences can be used to identify eating moments using two approaches: human computation, and by employing a combination of state-of-the-art machine learning and computer vision techniques. Furthermore, I also describe privacy challenges that arise with first-person photographs. Next, I present results showing how certain sounds associated with eating can be recognized and used to infer eating activities. Finally, I elaborate on findings from three studies focused on the use of on-body inertial sensors (head and wrists) to recognize eating moments both in a semi-controlled laboratory setting and in real-world conditions. I conclude by relating findings and insights to practical applications, and highlighting opportunities for future work.

abstract: Sarcasm is a nuanced form of language where usually, the speaker explicitly states the opposite of what is implied. Imbued with intentional ambiguity and subtlety, detecting sarcasm is a difficult task, even for humans. Current works approach this challenging problem primarily from a linguistic perspective, focusing on the lexical and syntactic aspects of sarcasm. In this thesis, I explore the possibility of using behavior traits intrinsic to users of sarcasm to detect sarcastic tweets. First, I theorize the core forms of sarcasm using findings from the psychological and behavioral sciences, and some observations on Twitter users. Then, I develop computational features to model the manifestations of these forms of sarcasm using the user's profile information and tweets. Finally, I combine these features to train a supervised learning model to detect sarcastic tweets. I perform experiments to extensively evaluate the proposed behavior modeling approach and compare with the state-of-the-art. / Dissertation/Thesis / Masters Thesis Computer Science 2014

Linux containers are gaining increasing traction in both individual and industrial use. As these containers get integrated into mission-critical systems, real-time detection of malicious cyber attacks becomes a critical operational requirement. However, a little research has been conducted in this area. This research introduces an anomaly-based intrusion detection and remediation system for container-based clouds. The introduced system monitors system calls between the container and the host server to passively detect malfeasance against applications running in cloud containers. We started by applying a basic memory-based machine learning technique to model the container behavior. The same technique was also extended to learn the behavior of a distributed application running in a number of cloud-based containers. In addition to monitoring the behavior of each container independently, the system used prior knowledge for a more informed detection system. We then studied the feasibility and effectiveness of applying a more sophisticated deep learning technique to the same problem. We used a recurrent neural network to model the container behavior. We evaluated the system using a typical web application hosted in two containers, one for the front-end web server, and one for the back-end database server. The system has shown promising results for both of the machine learning techniques used. Finally, we describe a number of incident handling and remediation techniques to be applied upon attack detection. / Ph. D.