Computer Sciences Commons^™

The Dangers Of Human-Like Bias In Machine-Learning Algorithms, Daniel James Fuchs

Missouri S&T’s Peer to Peer

Machine learning algorithms (MLAGs), frequently used in artificial intelligence (AI), rely on using patterns across sets of data to derive decision-making intelligence. In recent years, as society continues to give increasing authority to ML-driven AIs, these algorithms have demonstrated the ability to take on human-like discriminatory biases. Microsoft's "Tay," for example, a social media-based chatbot, went from resembling a normal teenage girl to displaying racist and sexist attitudes in a mere sixteen hours). Tay and many other ML-driven implementations across a wide variety of fields have replicated numerous human biases. In most cases, these human-like biases originated due to improper …

Software Engineering: Guest Editor's Introduction, Bruce M. Mcmillin

Computer Science Faculty Research & Creative Works

Engineering complex systems requires an extensive technical and semantic knowledge. Because of their complexity, components of these systems can interact in unpredictable ways -- sequentially and concurrently….

The three papers in this theme issue have the potential to address the design challenges of complex systems spanning multiple domains and interacting in potentially unexpected ways. The knowledge required goes beyond basic computing concepts and design, to include an examination of interactions and multiple aspects of the application domain. We hope you find these articles interesting and insightful.

A Group-Based Energy Harvesting Mac Protocol With Ap Scheduling In Machine-To-Machine Networks, Ce Zhou, Yunmin Kim, Tae-Jin Lee

Computer Science Faculty Research & Creative Works

In Machine-to-Machine (M2M) communications, an enormous number of devices are subject to the battery limitation. Recently, an attractive energy harvesting technology called energy beamforming has shown great potential to make wireless devices self-powering. Motivated by this, according to the IEEE 802.11ah, a group-based energy harvesting Medium Access Control (MAC) protocol with the strategy of AP scheduling is designed in the paper to address the energy shortage without much reduction in system throughput. In the proposed protocol, nodes first contend for the transmission opportunities, and then data and energy transfer simultaneously by the AP scheduling scheme. Multiple Power Beacons (PBs) are …

Nupt St-Data Miner: An Spatio-Temporal Data Analysis And Visualization System, Zhiqiang Zou, Junjie Xiong, Xuerong He, Haihong Dai

Computer Science Faculty Research & Creative Works

Given the increasing popularity and availability of location tracking devices, large quantities of Spatio-Temporal data (ST-data) are available from many different sources. For the ST-data, reflecting the mobile characteristic of the world, it is essential to build a functional system to perform quickly interactive analysis. In this paper, we present an analysis and visualization system, NUPT ST-data Miner, which facilitates users to visualize and analyze ST-data. It (1) provides a flexible and extensible framework based on cloud computing platform, (2) is able to quickly retrieve specified ST-data, (3) integrated multiple functions for the ST-data. To demonstrate its efficiency, we validate …

Mixed-Criticality Real-Time Task Scheduling With Graceful Degradation, Samsil Arefin

Masters Theses

”The mixed-criticality real-time systems implement functionalities of different degrees of importance (or criticalities) upon a shared platform. In traditional mixed-criticality systems, under a hi mode switch, no guaranteed service is provided to lo-criticality tasks. After a mode switch, only hi-criticality tasks are considered for execution while no guarantee is made to the lo-criticality tasks. However, with careful optimistic design, a certain degree of service guarantee can be provided to lo-criticality tasks upon a mode switch. This concept is broadly known as graceful degradation. Guaranteed graceful degradation provides a better quality of service as well as it utilizes the system resource …

Multitask Allocation To Heterogeneous Participants In Mobile Crowd Sensing, Weiping Zhu, Wenzhong Guo, Zhiyong Yu, Haoyi Xiong

Computer Science Faculty Research & Creative Works

Task allocation is a key problem in Mobile Crowd Sensing (MCS). Prior works have mainly assumed that participants can complete tasks once they arrive at the location of tasks. However, this assumption may lead to poor reliability in sensing data because the heterogeneity among participants is disregarded. In this study, we investigate a multitask allocation problem that considers the heterogeneity of participants (i.e., different participants carry various devices and accomplish different tasks). A greedy discrete particle swarm optimization with genetic algorithm operation is proposed in this study to address the abovementioned problem. This study is aimed at maximizing the number …

Containment Control Of Heterogeneous Systems With Non-Autonomous Leaders: A Distributed Optimal Model Reference Approach, Yongliang Yang, Shusen Cheng, Yixin Yin, Donald C. Wunsch

Electrical and Computer Engineering Faculty Research & Creative Works

This paper presents a distributed optimal model reference adaptive control approach for solving containment control of heterogeneous multi-agent systems (MASs) with non-autonomous leaders. First, a fully distributed adaptive observer is designed to provide for each agent the desired reference trajectory by estimating the convex hull spanned by leaders. The distributed observer dynamics serves as a reference model for each follower to synchronize. The global communication graph information or the leader dynamics is not required to design the observer. In contrast to existing model reference adaptive controllers (MRAC) for single-agent systems and containment control solutions for MASs, the proposed MRAC approach …

Learning From Experience: An Automatic Ph Neutralization System Using Hybrid Fuzzy System And Neural Network, Ethar H.K. Alkamil, Seaar Al-Dabooni, Ahmed K. Abbas, Ralph Flori, Donald C. Wunsch

Geosciences and Geological and Petroleum Engineering Faculty Research & Creative Works

In oil and gas industry, the pH level is one of the most important indicators of mud contamination while drilling. Although the process has simple components, the pH neutralization process is complicated in the mud circulation system. This difficulty is due to the high nonlinearity of the process. In this paper, the fuzzy neural network (FNN) is integrated to the fuzzy logic controller (FLC) to create a system which identifies pH fluctuation into a drilling mud system, assesses and signals its severity. And then, it automatically actuates a chemical treating fluids valve (CTFV), such as Caustic Soda (CS), for neutralizing …

Analyzing Sensor Based Human Activity Data Using Time Series Segmentation To Determine Sleep Duration, Yogesh Deepak Lad

Masters Theses

"Sleep is the most important thing to rest our brain and body. A lack of sleep has adverse effects on overall personal health and may lead to a variety of health disorders. According to Data from the Center for disease control and prevention in the United States of America, there is a formidable increase in the number of people suffering from sleep disorders like insomnia, sleep apnea, hypersomnia and many more. Sleep disorders can be avoided by assessing an individual's activity over a period of time to determine the sleep pattern and duration. The sleep pattern and duration can be …

Mining And Analysis Of Real-World Graphs, Armita Abedijaberi

Doctoral Dissertations

"Networked systems are everywhere - such as the Internet, social networks, biological networks, transportation networks, power grid networks, etc. They can be very large yet enormously complex. They can contain a lot of information, either open and transparent or under the cover and coded. Such real-world systems can be modeled using graphs and be mined and analyzed through the lens of network analysis. Network analysis can be applied in recognition of frequent patterns among the connected components in a large graph, such as social networks, where visual analysis is almost impossible. Frequent patterns illuminate statistically important subgraphs that are usually …

Reputation And Credit Based Incentive Mechanism For Data-Centric Message Delivery In Delay Tolerant Networks, Himanshu Jethawa

Masters Theses

"In a Data-centric Delay Tolerant Networks (DTNs), it is essential for nodes to cooperate in message forwarding in order to enable successful delivery of a message in an opportunistic fashion with nodes having their social interests defined. In the data-centric dissemination protocol proposed here, a source annotates messages (images) with keywords, and then intermediate nodes are presented with an option of adding keyword-based annotations in order to create higher content strength messages on path toward the destination. Hence, contents like images get enriched as there is situation evolution or learned by these intermediate nodes, such as in a battlefield, or …

Solidification Rate Detection Through Solid-Liquid Interface Tracking, Wei Luo

Masters Theses

”Studying the melt pool dynamics in laser powder bed fusion (LPBF) additive manufacturing is vital to predict the microstructure of the built part. High-speed synchrotron hard X-ray imaging is an advanced technique to monitor the LPBF process in situ and in real time. However, it is very challenging to track the solid-liquid interface due to the low image contrast, high image noise and unstable intensity of the image sequences.

In this paper, we propose an solid-liquid interface detector with human interaction to track the interface to compute the solidification rate of the melt pool. The proposed method includes six independent …

Multiple Security Domain Non Deducibility In The Freedm Smart Grid Infrastructure, Manish Jaisinghani

Masters Theses

"The building block of today's world are not materials, but, computers and algorithms with communication networks between physical entities. A cyber physical system (CPS) is a system in which the cyber and physical entities of the system work together towards a common goal, for example a water treatment facility or an electricity distribution system. These cyber physical infrastructures affect day to day lives of people and hence become target point for the attackers to disrupt normal daily life. Owing to the complexity of a cyber physical system, the attacks have themselves become sophisticated and harder to detect. These sophisticated attacks …

Cloud Transactions And Caching For Improved Performance In Clouds And Dtns, Dileep Mardham

Masters Theses

"In distributed transactional systems deployed over some massively decentralized cloud servers, access policies are typically replicated. Interdependencies ad inconsistencies among policies need to be addressed as they can affect performance, throughput and accuracy. Several stringent levels of policy consistency constraints and enforcement approaches to guarantee the trustworthiness of transactions on cloud servers are proposed. We define a look-up table to store policy versions and the concept of "Tree-Based Consistency" approach to maintain a tree structure of the servers. By integrating look-up table and the consistency tree based approach, we propose an enhanced version of Two-phase validation commit (2PVC) protocol integrated …

Cyber-Physical Security Of An Electric Microgrid, Prashanth Palaniswamy

Masters Theses

"Cyber-physical systems (CPSs) are physical systems that are controlled or monitored by computer-based systems. CPSs are a combination of computation, networking, and physical processes. As CPSs are a combination of various diverse components, they are vulnerable to several security threats. Moreover, there are many different security domains (not just high/low, nor necessarily hierarchical). This paper utilizes previously developed multiple security domain nondeducibility (MSDND) to uncover potential integrity vulnerabilities in an electric microgrid. Invariants are manually generated using the insights obtained through MSDND analysis and use linear regression to automate the generation of invariants. The vulnerabilities are then mitigated, to the …

An Approach For Formal Analysis Of The Security Of A Water Treatment Testbed, Sai Sidharth Patlolla

Masters Theses

"This thesis focuses on securing critical infrastructures such as chemical plants, manufacturing units, and power generating plants against attacks that disrupt the information flow from one component to another. Such systems are controlled by an Industrial Control System (ICS) that includes controllers communicating with each other, and with physical sensors and actuators, using a communications network.

Traditional security models partition the security universe into two worlds, secure and insecure, but in the real world the partitions often overlap and information is leaked even through the physical observation which makes it much harder to analyze a Cyber physical system (CPS). To …

A Network Tomography Approach For Traffic Monitoring In Smart Cities, Ruoxi Zhang

Masters Theses

"Various urban planning and managing activities required by a Smart City are feasible because of traffic monitoring. As such, the thesis proposes a network tomography-based approach that can be applied to road networks to achieve a cost-efficient, flexible, and scalable monitor deployment. Due to the algebraic approach of network tomography, the selection of monitoring intersections can be solved through the use of matrices, with its rows representing paths between two intersections, and its columns representing links in the road network. Because the goal of the algorithm is to provide a cost-efficient, minimum error, and high coverage monitor set, this problem …

Card: Concealed And Remote Discovery Of Iot Devices In Victims' Home Networks, Sammie Lee Bush

Masters Theses

"Smart devices are becoming more common in the standard households. They range from lights to refrigerators and their functionality and applications continues to grow with consumer demand. This increase in networked, complex devices has also brought an increase in vulnerabilities in the average consumer's home. There now exists an Internet of Things (IoT) ecosystem that creates new attack vectors for adversaries to spread malware, build botnets, and participate in other malicious activities. We will overview some of these new attack vectors as well as go over a framework that would allow an adversary to target a user's home network and …

Analyzing Large Scale Trajectory Data To Identify Users With Similar Behavior, Tyler Clark Percy

Masters Theses

"In today's society, social networks are a popular way to connect with friends and family and share what's going on in your life. With the Internet connecting us all closer than ever before, it is increasingly common to use social networks to meet new friends online that share similar interests instead of only connecting with those you already know. For the problem of attempting to connect people with similar interests, this paper proposes the foundation for a Geo-social network that aims to extract the semantic meaning from users' location history and use this information to find the similarity between users. …

Precise Energy Efficient Scheduling Of Mixed-Criticality Tasks & Sustainable Mixed-Criticality Scheduling, Sai Sruti

Masters Theses

"In this thesis, the imprecise mixed-criticality model (IMC) is extended to precise scheduling of tasks, and integrated with the dynamic voltage and frequency scaling (DVFS) technique to enable energy minimization. The challenge in precise scheduling of MC systems is to simultaneously guarantee the timing correctness for all tasks, hi and lo, under both pessimistic and optimistic (less pessimistic) assumptions. To the best of knowledge this is the first work to address the integration of DVFS energy conserving techniques with precise scheduling of lo-tasks of the MC model.

In this thesis, the utilization based schedulability tests and sufficient conditions for such …

Smart Augmented Reality Instructional System For Mechanical Assembly, Ze-Hao Lai

Masters Theses

"Quality and efficiency are pivotal indicators of a manufacturing company. Many companies are suffering from shortage of experienced workers across the production line to perform complex assembly tasks such as assembly of an aircraft engine. This could lead to a significant financial loss. In order to further reduce time and error in an assembly, a smart system consisting of multi-modal Augmented Reality (AR) instructions with the support of a deep learning network for tool detection is introduced. The multi-modal smart AR is designed to provide on-site information including various visual renderings with a fine-tuned Region-based Convolutional Neural Network, which is …

Classification Of Eeg Signals Of User States In Gaming Using Machine Learning, Chandana Mallapragada

Masters Theses

"In this research, brain activity of user states was analyzed using machine learning algorithms. When a user interacts with a computer-based system including playing computer games like Tetris, he or she may experience user states such as boredom, flow, and anxiety. The purpose of this research is to apply machine learning models to Electroencephalogram (EEG) signals of three user states -- boredom, flow and anxiety -- to identify and classify the EEG correlates for these user states. We focus on three research questions: (i) How well do machine learning models like support vector machine, random forests, multinomial logistic regression, and …

Improved Crpd Analysis And A Secure Scheduler Against Information Leakage In Real-Time Systems, Ying Zhang

Masters Theses

"Real-time systems are widely applied to the time-critical fields. In order to guarantee that all tasks can be completed on time, predictability becomes a necessary factor when designing a real-time system. Due to more and more requirements about the performance in the real-time embedded system, the cache memory is introduced to the real-time embedded systems.

However, the cache behavior is difficult to predict since the data will be loaded either on the cache or the memory. In order to taking the unexpected overhead, execution time are often enlarged by a certain (huge) factor. However, this will cause a waste of …

Machine Learning Techniques Implementation In Power Optimization, Data Processing, And Bio-Medical Applications, Khalid Khairullah Mezied Al-Jabery

Doctoral Dissertations

"The rapid progress and development in machine-learning algorithms becomes a key factor in determining the future of humanity. These algorithms and techniques were utilized to solve a wide spectrum of problems extended from data mining and knowledge discovery to unsupervised learning and optimization. This dissertation consists of two study areas. The first area investigates the use of reinforcement learning and adaptive critic design algorithms in the field of power grid control. The second area in this dissertation, consisting of three papers, focuses on developing and applying clustering algorithms on biomedical data. The first paper presents a novel modelling approach for …

Mitigation Of Environmental Hazards Of Sulfide Mineral Flotation With An Insight Into Froth Stability And Flotation Performance, Muhammad Badar Hayat

Doctoral Dissertations

"Today's major challenges facing the flotation of sulfide minerals involve constant variability in the ore composition; environmental concerns; water scarcity and inefficient plant performance. The present work addresses these challenges faced by the flotation process of complex sulfide ore of Mississippi Valley type with an insight into the froth stability and the flotation performance. The first project in this study was aimed at finding the optimum conditions for the bulk flotation of galena (PbS) and chalcopyrite (CuFeS₂) through Response Surface Methodology (RSM). In the second project, an attempt was made to replace toxic sodium cyanide (NaCN) with the biodegradable chitosan …

Dragline Excavation Simulation, Real-Time Terrain Recognition And Object Detection, Godfred Somua-Gyimah

Doctoral Dissertations

"The contribution of coal to global energy is expected to remain above 30% through 2030. Draglines are the preferred excavation equipment in most surface coal mines. Recently, studies toward dragline excavation efficiency have focused on two specific areas. The first area is dragline bucket studies, where the goal is to develop new designs which perform better than conventional buckets. Drawbacks in the current approach include operator inconsistencies and the inability to physically test every proposed design. Previous simulation models used Distinct Element Methods (DEM) but they over-predict excavation forces by 300% to 500%. In this study, a DEM-based simulation model …

Detecting Cells And Analyzing Their Behaviors In Microscopy Images Using Deep Neural Networks, Yunxiang Mao

Doctoral Dissertations

"The computer-aided analysis in the medical imaging field has attracted a lot of attention for the past decade. The goal of computer-vision based medical image analysis is to provide automated tools to relieve the burden of human experts such as radiologists and physicians. More specifically, these computer-aided methods are to help identify, classify and quantify patterns in medical images. Recent advances in machine learning, more specifically, in the way of deep learning, have made a big leap to boost the performance of various medical applications. The fundamental core of these advances is exploiting hierarchical feature representations by various deep learning …

Security Risk Assessment In Cloud Computing Domains, Amartya Sen

Doctoral Dissertations

"Cyber security is one of the primary concerns persistent across any computing platform. While addressing the apprehensions about security risks, an infinite amount of resources cannot be invested in mitigation measures since organizations operate under budgetary constraints. Therefore the task of performing security risk assessment is imperative to designing optimal mitigation measures, as it provides insight about the strengths and weaknesses of different assets affiliated to a computing platform.

The objective of the research presented in this dissertation is to improve upon existing risk assessment frameworks and guidelines associated to different key assets of Cloud computing domains - infrastructure, applications, …

Tensorviz: Visualizing The Training Of Convolutional Neural Network Using Paraview, Xinyu Chen, Qiang Guan, Xin Liang, Li-Ta Lo, Simon Su, Trilce Estrada, James Ahrens

Computer Science Faculty Research & Creative Works

Deep Convolutional Networks have been very successful in visual recognition tasks recently. Previous works visualize learned features at different layers o help people to understand how CNNs learn visual recognition tasks. However they do not help to accelerate the training process. We use Paraview to provides both qualitative and quantitative visualization that help understand the learning procedure, tune the learning parameters and direct merging and pruning of neural networks.

Tensorviz: Visualizing The Training Of Convolutional Neural Network Using Paraview (Poster), Xinyu Chen, Qiang Guan, Xin Liang, Li-Ta Lo, Simon Su, Trilce Estrada, James Ahrens

Computer Science Faculty Research & Creative Works

Deep Convolutional Networks have been very successful in visual recognition tasks recently. Lots of previous works aimed to help people to get senses of why those biology-inspired networks achieved such good performances. Deconvnet[1], Guided propagation[2] and a comprehensive visualization tool box[3] can help people to see features learned at different layers of the networks. These works in some extent provided understanding and support for the biology origin of how convolutional networks emulate visual recognition tasks. However, due to the complexity of searching in very high dimensional parameter space, the whole training remains in black-boxes. Normally a large network needs weeks …