Computer Engineering Commons^™

Reinforcement Learning In Robotic Task Domains With Deictic Descriptor Representation, Harry Paul Moore

LSU Doctoral Dissertations

In the field of reinforcement learning, robot task learning in a specific environment with a Markov decision process backdrop has seen much success. But, extending these results to learning a task for an environment domain has not been as fruitful, even for advanced methodologies such as relational reinforcement learning. In our research into robot learning in environment domains, we utilize a form of deictic representation for the robot’s description of the task environment. However, the non-Markovian nature of the deictic representation leads to perceptual aliasing and conflicting actions, invalidating standard reinforcement learning algorithms. To circumvent this difficulty, several past research …

Effective Methods And Tools For Mining App Store Reviews, Nishant Jha

LSU Doctoral Dissertations

Research on mining user reviews in mobile application (app) stores has noticeably advanced in the past few years. The main objective is to extract useful information that app developers can use to build more sustainable apps. In general, existing research on app store mining can be classified into three genres: classification of user feedback into different types of software maintenance requests (e.g., bug reports and feature requests), building practical tools that are readily available for developers to use, and proposing visions for enhanced mobile app stores that integrate multiple sources of user feedback to ensure app survivability. Despite these major …

Bipartite Quantum Interactions: Entangling And Information Processing Abilities, Siddhartha Das

LSU Doctoral Dissertations

The aim of this thesis is to advance the theory behind quantum information processing tasks, by deriving fundamental limits on bipartite quantum interactions and dynamics. A bipartite quantum interaction corresponds to an underlying Hamiltonian that governs the physical transformation of a two-body open quantum system. Under such an interaction, the physical transformation of a bipartite quantum system is considered in the presence of a bath, which may be inaccessible to an observer. The goal is to determine entangling abilities of such arbitrary bipartite quantum interactions. Doing so provides fundamental limitations on information processing tasks, including entanglement distillation and secret key …

Heuristics For Client Assignment And Load Balancing Problems In Online Games, Shawn Michael Farlow

LSU Doctoral Dissertations

Massively multiplayer online games (MMOGs) have been very popular over the past decade. The infrastructure necessary to support a large number of players simultaneously playing these games raises interesting problems to solve. Since the computations involved in solving those problems need to be done while the game is being played, they should not be so expensive that they cause any noticeable slowdown, as this would lead to a poor player perception of the game. Many of the problems in MMOGs are NP-Hard or NP-Complete, therefore we must develop heuristics for those problems without negatively affecting the player experience as a …

A Study Of Scalability And Cost-Effectiveness Of Large-Scale Scientific Applications Over Heterogeneous Computing Environment, Arghya K. Das

LSU Doctoral Dissertations

Recent advances in large-scale experimental facilities ushered in an era of data-driven science. These large-scale data increase the opportunity to answer many fundamental questions in basic science. However, these data pose new challenges to the scientific community in terms of their optimal processing and transfer. Consequently, scientists are in dire need of robust high performance computing (HPC) solutions that can scale with terabytes of data.

In this thesis, I address the challenges in three major aspects of scientific big data processing as follows: 1) Developing scalable software and algorithms for data- and compute-intensive scientific applications. 2) Proposing new cluster architectures …

Distributed Iterative Graph Processing Using Nosql With Data Locality, Ayam Pokhrel

LSU Master's Theses

A tremendous amount of data is generated every day from a wide range of sources such as social networks, sensors, and application logs. Among them, graph data is one type that represents valuable relationships between various entities. Analytics of large graphs has become an essential part of business processes and scientific studies because it leads to deep and meaningful insights into the related domain based on the connections between various entities. However, the optimal processing of large-scale iterative graph computations is very challenging due to the issues like fault tolerance, high memory requirement, parallelization, and scalability. Most of the contemporary …

Resampling Methods And Visualization Tools For Computer Performance Comparisons In The Presence Of Performance Variation, Samuel Oridge Irving

LSU Master's Theses

Performance variability, stemming from non-deterministic hardware and software behaviors or deterministic behaviors such as measurement bias, is a well-known phenomenon of computer systems which increases the difficulty of comparing computer performance metrics and is slated to become even more of a concern as interest in Big Data Analytics increases. Conventional methods use various measures (such as geometric mean) to quantify the performance of different benchmarks to compare computers without considering this variability which may lead to wrong conclusions. In this paper, we propose three resampling methods for performance evaluation and comparison: a randomization test for a general performance comparison between …

Image Processing Applications In Real Life: 2d Fragmented Image And Document Reassembly And Frequency Division Multiplexed Imaging, Houman Kamran Habibkhani

LSU Doctoral Dissertations

In this era of modern technology, image processing is one the most studied disciplines of signal processing and its applications can be found in every aspect of our daily life. In this work three main applications for image processing has been studied.

In chapter 1, frequency division multiplexed imaging (FDMI), a novel idea in the field of computational photography, has been introduced. Using FDMI, multiple images are captured simultaneously in a single shot and can later be extracted from the multiplexed image. This is achieved by spatially modulating the images so that they are placed at different locations in the …

Evaluating Classifiers' Optimal Performances Over A Range Of Misclassification Costs By Using Cost-Sensitive Classification, Ramy Al-Saffar

LSU Master's Theses

We believe that using the classification accuracy is not enough to evaluate the performances of classification algorithms. It can be misleading due to overlooking an important element which is the cost if classification is inaccurate. Furthermore, the Receiver Operational Characteristic (ROC) is one of the most popular graphs used to evaluate classifiers performances. However, one of the biggest ROC’s shortcomings is the assumption of equal costs for all misclassified data. Therefore, our goal is to reduce the total cost of decision making by selecting the classifier that has the least total misclassification cost. Nevertheless, the exact misclassification cost is usually …