Engineering Commons^™

Teachability And Interpretability In Reinforcement Learning, Jeevan Rajagopal

School of Computing: Dissertations, Theses, and Student Research

There have been many recent advancements in the field of reinforcement learning, starting from the Deep Q Network playing various Atari 2600 games all the way to Google Deempind's Alphastar playing competitively in the game StarCraft. However, as the field challenges more complex environments, the current methods of training models and understanding their decision making become less effective. Currently, the problem is partially dealt with by simply adding more resources, but the need for a better solution remains.

This thesis proposes a reinforcement learning framework where a teacher or entity with domain knowledge of the task to complete can assist …

“The Revolution Will Not Be Supervised": An Investigation Of The Efficacy And Reasoning Process Of Self-Supervised Representations, Atharva Tendle

School of Computing: Dissertations, Theses, and Student Research

Transfer learning technique enables training Deep Learning (DL) models in a data-efficient way for solving computer vision tasks. It involves pretraining a DL model to learn representations from a large and general-purpose source dataset, then fine-tuning the model using the task-specific target dataset. The dominant supervised learning (SL) approach for pretraining representations suffers from some limitations that include expensive labeling and poor generalizability. Recent advancements in the self-supervised learning (SSL) approach made it possible to learn effective representations from unlabeled data. The performance of the fine-tuned DL models based on pretrained SSL representations is on par with the state-of-the-art pretrained …

Development Of A Software For The Optimization Of Mechanical Structural Elements, Joshua Gross

Honors Capstone Projects

The goal of this software is to develop the framework for the optimal design of a mechanical structural element that minimizes the total part volume while maintaining the induced stress below an acceptable level. The software starts with input from the user, which is a set of ANSYS APDL commands that include constant dimensions, applied force, and initial values for the variable dimensions. Then, a MATLAB script modifies the variable dimensions, and the output is analyzed by the nonlinear programming solver known as Sequential Gradient-Restoration Algorithm. Finally, the solver converges on the dimensions that produce a structurally viable element with …

Demo Abstract: Porting And Execution Of Anomalies Detection Models On Embedded Systems In Iot, Bharath Sudharsan, Pankesh Patel, Abdul Wahid, Muhammad Yahya, John G. Breslin, Muhammad Intizar Ali

Publications

In the Industry 4.0 era, Microcontrollers (MCUs) based tiny embedded sensor systems have become the sensing paradigm to interact with the physical world. In 2020, 25.6 billion MCUs were shipped, and over 250 billion MCUs are already operating in the wild. Such low-power, low-cost MCUs are being used as the brain to control diverse applications and soon will become the global digital nervous system. In an Industrial IoT setup, such tiny MCU-based embedded systems are equipped with anomaly detection models and mounted on production plant machines for monitoring the machine’s health/condition. These models process the machine’s health data (from temperature, …

An Investigation Of Runway Risks Using A Systems Approach, Mehnaz Hassan

Theses and Dissertations

The purpose of this study was to explore and determine if Causal Analysis Based on System Theory ( could divulge differing underlying causes in the Taxiway overflight incident of Air Canada Flight 759 at San Fran cisco International Airport, as compared to traditional investigation methods used by the National Transportation Safety Board ( Investigations tend to focus on trying to establish the probable cause of an accident or incident, and as a result, the systemic components tend to be overlooked which leads to the common underlying causes of the event left unsolved. Adopting a systems approach to assessing runway risks …

Computational Frameworks For Microrna Functional Analysis Of Inter-Kingdom And Indirect Targeting., Mohammed Sayed

Electronic Theses and Dissertations

Genes are DNA sequences that encode the information needed to synthesize molecules necessary for the function of the cell. Some genes are called protein-coding genes because they have the code required to manufacture proteins. The expression of a certain gene means its product (protein) is produced. Although some genes are not protein-coding, they regulate the gene expression of other protein-coding genes. Of these, microRNAs (miRNAs) are small RNA molecules that inhibit the expression of other genes by binding to their mRNA transcripts. miRNAs have been shown to be linked to several biological processes like development and diseases like cancer. Recently, …

Urban Arterial Lane Flow Distribution: A Before And After Traffic Microsimulation Analysis On The Effect Of Implementing New Route Markings And Signage., Larry Michael Summers

Electronic Theses and Dissertations

Urban arterials may not effectively utilize available lanes to move vehicles efficiently as drivers try to establish position along the corridor. This study evaluates lane flow distribution before and after implementing new route markings and overhead signs to utilize existing lanes better. A cellphone-based traffic data service called StreetLight gathered the traffic data used to establish the VISSIM model parameters for simulating current and projected conditions. This study shows that the lane assignment signage and route markers affected the lane flow distribution in the simulation. The simulation indicated the existing condition had average queue lengths of 320.6 meters, average travel …

Machine Learning Approaches For Lung Cancer Diagnosis., Ahmed Mahmoud Ahmed Shaffie

Electronic Theses and Dissertations

The enormity of changes and development in the field of medical imaging technology is hard to fathom, as it does not just represent the technique and process of constructing visual representations of the body from inside for medical analysis and to reveal the internal structure of different organs under the skin, but also it provides a noninvasive way for diagnosis of various disease and suggest an efficient ways to treat them. While data surrounding all of our lives are stored and collected to be ready for analysis by data scientists, medical images are considered a rich source that could provide …

Thermo-Hydro-Mechanical Characterization Of Ottawa Sand And Kaolin Clay Through Experimental Models., Mohammad Joshaghani

Electronic Theses and Dissertations

Recent advances in energy geotechnics and increased utilization of ground source heat pump systems (e.g., geothermal boreholes and energy y piles) and energy geo-structures require improving the understanding of the thermo-hydro-mechanical behavior of soil. Energy geo-structures consist of above-ground facilities, buried structures, and the soil surrounding the buried structure. A buried structure transfers the energy between the ground and above-ground facilities. Energy transfer between above-ground facilities and the soil will cause temperature changes in soil and this will affect soil hydro-mechanical properties. The changes in soil temperature can also significantly alter the soil-structure interaction and its mechanical strength. Soil temperature …

Restaurant Style Prediction Using Word2vec And Support Vector Machine., Saleh Abdullah Almohaimeed

Electronic Theses and Dissertations

Natural Language Processing represents a quantum leap for governance and industries. It enables them to have an insight into hidden patterns and information within their data. In this thesis, we have worked on an important field in Natural Language Processing, which is Text Classification. Our goal is to help restaurant owners to find which dishes customers like more. To do that we have used a dataset from Yelp.com that has 150,000 restaurant reviews, then count the most frequent dishes mentioned. However, this way is not effective except if these reviews are categorized into different restaurants-styles. For this reason, we have …

Effects Of Flow Path Factors In The Permeability Of Natural And Man-Made Granular Soils Using The Kozeny-Carman Equation, Robert Davies

All Graduate Plan B and other Reports, Spring 1920 to Spring 2023

The research presented in this paper serves to observe how the flow path factors relate to the void ratio, the effective diameter of a soil particle, and the permeability of a given sample assuming the behavior is consistent across natural and artificial soils. The basis of the comparison is the derived Kozeny-Carman Equation for permeability with focus on the factors C_s and C_l where C_s represents the shape of the flow path and C_l represents the length of the flow path that a single water molecule must travel through a soil sample. Permeameter tests were conducted …

Surface Engineering Of Cerium Oxide Nanocyrstal Dispersions: Colloidal Properties, Ageing Effects, & Electroanalysis, Craig Neal

Electronic Theses and Dissertations, 2020-2023

Colloidal materials are highly diverse and present complex physicochemical properties which define their utilities in applications spanning diverse industries. In particular, nano-scale colloids have received tremendous attention due to their unique, specific activities as compared to larger-sized preparations. Within the biomedical sciences, nano-colloids are routinely used as inert carriers of therapeutics and/or diagnostic agents. Beyond this, researchers have developed functional colloids which demonstrate bio-active or diagnostic properties themselves: often related to an optimized, or tuned, surface character. Among these, nanoscale cerium oxide (nanoceria) has shown great promise as a bi-functional, therapeutic material: producing pro- or anti- oxidative chemical response in …

Secondary Resonances Of Electrostatically Actuated Mems And Nems Circular Plates, Julio S. Beatriz

Theses and Dissertations

Two methods are used in the investigation, which are the Method of Multiple Scales (MMS) and the Reduced order Model (ROM) with up to 6 modes of vibration. The MMS is done in three ways, one based on soft excitation and the other models based on the usage of the bookkeeping parameter, ε, in the electrostatic term. These are referred to as the 1 term no ε, and the 2 term no ε, which allow it to model hard excitation. The influence of the number of terms in the Taylor expansion is explored.

These methods are used to investigate the …

Characterization Of The Rf Magnetron Sputtered P-Si & N-Si On Si Substrate And Effect Of Process Parameters On Their Structural & Electrical Properties, Prosanto Biswas

Theses and Dissertations

In our work, we applied Taguchi Signal-to-noise (SNR) analysis to investigate the effect of varying three process parameters, namely- sputtering power, working pressure, and Ar gas flow rate on the surface and morphological properties of the RF sputtered p-Si and n-Si on Si substrate. We also inspected the contribution of process parameters on those properties by applying Analysis of Variance (ANOVA). Characteristics of thin films fabricated by RF magnetron sputtering vary with the recipes. Process parameters such as power, pressure, process gas flow rate, substrate temperature, and target to substrate distance determine the quality of the thin films. Some of …

Novel Coating Methods On Centrifugally-Spun Polymer Fibers For Applications In Lithium-Ion Batteries, Jonathan G. Ayala

Theses and Dissertations

The work presented in this thesis focuses on the processing, characterization, and electrochemical results of centrifugally spun composite carbon fiber electrodes for application as anode material in lithium-ion batteries. The work is presented as a compilation of two major projects. First, the use of novel Co3O4 wet coatings to increase the capacity of carbon fibers produced from Polyacrylonitrile (PAN). In this work, PAN fibers are produced via the Forcespinning method, and were heat treated by oxidation in air at 200 °C for four hours, and subsequent carbonization at 600 °C for six hours. The electrochemical performance of the Co3O4/C composite-fiber …

Knee Angles After Crosstalk Correction With Principal Component Analysis In Gait And Cycling, Jordan Skaro, Scott J. Hazelwood, Stephen M. Klisch

Biomedical Engineering

Principal component analysis (PCA) has been used as a post-hoc method for reducing knee crosstalk errors during gait analysis. PCA minimizes correlations between flexion–extension (FE), abduction–adduction (AA), and internal–external rotation (IE) angles. However, previous studies have not considered PCA for exercises involving knee flexion angles that are greater than those typically experienced during gait. Thus, the goal of this study was to investigate using PCA to correct for crosstalk during one exercise (i.e., cycling) that involves relatively high flexion angles. Fifteen participants were tested in gait and cycling using a motion analysis system. Uncorrected FE, AA and IE angles were …

Impact Of Wind Generation On Transmission Voltage Support During Weak Grid, Yugo Isogai

Electrical Engineering Undergraduate Honors Theses

Today, there are concerns about the effects of high wind penetration on electric power system operations. This is due to the more distributed nature of wind turbine generation and the corresponding multi-path power flows on the electric power transmission grid. In particular, the higher line impendences associated with inverter-based wind generation results in larger voltage fluctuation under varying load. This may results in conditions where even after a fault condition has been cleared, the transmission system voltages may not recover. Therefore, this paper concentrates on the impact of weak grid condition for voltage instability. To approach it, a 6-bus system …

Process Design And Optimization Of A Once-Through Dimethyl Ether Process Using Aveva Process Simulation, Jerrod Scott

Honors Theses

The primary purpose of the project on which I wrote this thesis was to design an optimize a once-through dimethyl ether process utilizing a software called AVEVA Process Simulation. The process was initially modeled and optimized to minimize the EAOC of the distillation column. The result of that effort was an optimized column with six stages with the feed entering the column at stage six. It had a height of 3 m, diameter of 0.44 m, and an EAOC of $65,100. This optimized solution was submitted to a Toller. The Toller provided available equipment to model the plant. From here, …

Immunomodulatory Biomaterials For Cancer Immunotherapy, Larry Donnell Stokes Jr

Honors Theses

Cancer immunotherapy has become an effective treatment in the toolbox of oncologists. Immunotherapy offers a less toxic alternative to standard cancer treatments such as chemotherapy and can have prolonged curative effects to decrease cancer recurrence. Today, many drugs and biological agents have been developed that target the immune system and elicit an antitumor/cancer response. These agents are known collectively as cancer immunotherapies. While immunotherapies have radically improved treatment outcomes for many cancer patients, there are drawbacks to using these treatments. Immunotherapy treatments have poor clinical responses in patients with tumors that lack immunogenicity. Some of the treatments also pose a …

A Study Of Deep Reinforcement Learning In Autonomous Racing Using Deepracer Car, Mukesh Ghimire

Honors Theses

Reinforcement learning is thought to be a promising branch of machine learning that has the potential to help us develop an Artificial General Intelligence (AGI) machine. Among the machine learning algorithms, primarily, supervised, semi supervised, unsupervised and reinforcement learning, reinforcement learning is different in a sense that it explores the environment without prior knowledge, and determines the optimal action. This study attempts to understand the concept behind reinforcement learning, the mathematics behind it and see it in action by deploying the trained model in Amazon's DeepRacer car. DeepRacer, a 1/18^th scaled autonomous car, is the agent which is trained …

Economic Feasibility Of A Methanol To Dimethyl Ether Production Process To Avoid Contract Failure Shortfalls From The Covid-19 Pandemic, Jacob Noll, Robert Wasson, Harrison Mckinnis

Honors Theses

Our team entered the 2021 AVEVA Academic Competition, where teams of undergraduate senior chemical engineering students competed across the country. The competition was composed of two parts: the base case design and the optimization of a chemical process. As part of the competition, our team is acting as the Engineering team for a fictional company that has given us this project. Due to COVID-19, our methanol producing company has lost a contract with a customer, leaving 23,000 tonnes/yr of unclaimed methanol. We have two choices with this methanol: either sell the methanol on the market at the spot price for …

Design Of An Undergraduate Laboratory Experiment Utilizing A Stirred-Tank, Jacketed Bioreactor, Brennen Middleton

Honors Theses

Bioreactors are utilized in many industries, such as the food, alternative fuel, and pharmaceutical industries, to design and manufacture products. Unlike the similar chemical batch reactors utilized in many chemical processing facilities, bioreactor utilization requires engineers to consider a wider range of operating conditions and parameters. This results in a more complex reaction system and controls network. Due to this, it is imperative for chemical and biomedical engineering students to not only understand the theory surrounding these reactor systems, but also understand how to properly design and perform operating procedures with these systems. Thus, it was determined to create an …

Ethylbenzene Production Case Study And Optimization, Brandon Borek

Honors Theses

The objective of this project was to investigate the process of producing ethylbenzene from the liquid-phase reaction of ethylene and benzene and perform parametric and topological optimizations to increase the net present value of the production facility. To do so, the base case conditions of the process were modeled and quantified economically followed by an investigation into the existing process to determine conditions that could be altered to improve the economic viability of the project. After optimizing the parametric conditions and topology of the process, the net present value of the project was increased by 305 million dollars to 191 …

Evaluation And Sizing Of Solar Powered Reverse Osmosis Water Desalination System, Sandip Subedi

Honors Theses

About 71 % of the surface of Earth is water, out of which around 2.5 % is freshwater. The majority of the surface of Earth is occupied by seawater which also contains dissolved salt. The salt concentration in seawater makes it toxic to humans because the human body is unable to get rid of the salt that comes from seawater. The demand for fresh water is increasing due to the rise in global population. In this context, desalinated seawater is considered as a viable resource to meet the growing demand for fresh water. Desalination process is highly energy intensive due …

Analysis On Fuel Options For Scramjet Engines With The Study To Lower The Starting Mach Number, Sumantra Luitel, Gagan Dangi

Honors Theses

The main objectives of this report were to perform analysis of an ideal scramjet engine, to assess the influence of fuel on endurance factor, and the possibility of lowering the starting Mach Number of the scramjet. In the first part, an ideal cycle parametric analysis was conducted on three different fuels i.e. Liquid Hydrogen (LH2), Jet Propellant 7 (JP-7), and Rocket Propellant (RP-1), taking into account their availability, physical properties, current uses, and potential uses. The detailed analysis is done largely relying on a 9 step Parametric Cycle Analysis technique to study how fuel properties influence the variation of seven …

Evaluating The Effects Of Seismic Loading On Concrete Reinforced Versus Precast Concrete Slab Floor Systems, Mahima Maharjan

Honors Theses

This thesis is part of the senior design capstone project of Ole Miss Department of Civil Engineering. The seismic loadings for the commercial buildings project located in Oxford, Mississippi are evaluated to fulfill the honor thesis requirement. This thesis follows the codes from American Concrete Institute (ACI), International Building Code (IBC), and ASCE-7/SEI (Minimum Design Loads and Associated Criteria for Buildings and Other Structures/Structural Engineering Institute) to design the gravitational loadings and seismic loadings.

Furthermore, a reinforced concrete slab flooring system is replaced with a precast slab flooring system to analyze the most seismic performance effective floor system for the …

Design And Production Of A Bathtub Caddy, John Marquez

Honors Theses

The purpose of this thesis is to document and discuss the design and manufacturing process used to produce a bathtub caddy on a large-scale production process for the Center for Manufacturing Excellence Senior Capstone Project. This project was carried out over the Fall of 2020 and Spring 2021 semesters by a team consisting of an Undergraduate Chemical Engineering Undergraduate, two Mechanical Engineering Undergraduate, a Finance Undergraduate, and an Accounting Undergraduate. The objective of this project was to design the product and then a manufacturing process that could be used to at first produce a prototype before planning a ramp up …

Design Of Load Bearing Wall For Low Rise Building With Partially Grouted Reinforced Masonry, Anil Bhatt

Honors Theses

The seismic and wind load acting on the 2-storeyed building of dimension 120 ft x 98 ft located in Oxford, MS, were calculated and the seismic load was considered for the design of the 120 ft long and 24 ft high load-bearing wall because it being critical. The maximum loading was computed using different load combinations. The masonry behavior and masonry specifications were considered to select the masonry unit, grout, and mortar for the load-bearing wall. The seismic design requirement for the shear and slender wall was fulfilled for the special reinforced masonry wall. The in-plane and out-of-plane loading scenarios …

Standardizing Patient Generated Sleep And Step Data For Evidation Health, Jade Desmet, Monroe Erle, Jamie Cullen

Industrial and Manufacturing Engineering

Since medical data has gone digital there has been a lack of standardization among wearable devices for managing the data. Evidation Health is a company that collects biometric data from a wide variety of wearable devices through their app, Achievement. Evidation is built upon a foundation of user privacy and control over permissioned health data. This project is in partnership with them, focusing on creating a standard data model and ETL tool for sleep and step data across the wearable devices Fitbit, Apple Watch, and Oura Ring. There is not yet a successful standardization for this health data, and the …

Control System For Variable-Orifice Mechatronic Spray Nozzle For Regulation Of Application Rate And Droplet Size Spectra, Rodney Rohrer

Department of Agricultural and Biological Systems Engineering: Dissertations, Theses, and Student Research

Traditional fixed-orifice spray nozzles are selected for flow rate and droplet spectra required for a given pesticide application. Although limited variation in flow can be achieved by adjusting system pressure, this can adversely affect spray quality. Other nozzle configurations, such as Pulse Width Modulated (PWM) nozzles or passive variable-orifice designs, are intended to maintain consistent droplet size and spray pattern as flow rate is changed but those too have limitations.

A variable-orifice nozzle modified by Luck (2012) can span multiple droplet spectra and offers independent control of flow and droplet size; however, an advanced control system must be developed to …