Texas A&M University-Kingsville: AKM Digital Repository
Not a member yet
1689 research outputs found
Sort by
Prediction of drug-potent proteins using deep learning
No full textIdentifying and locating drug targets is highly important for drug design and discovery. In contrast to the existing time-consuming conventional method that relies on examining the three-dimensional structure of the protein, an artificial intelligence algorithm can be utilized to predict a drug target effectively and efficiently directly from the protein sequence. In this research, features are extracted from a protein sequence using three different feature extraction and encoding mechanisms, which include: i) pseudo amino acid composition (PseAAC); ii) reduced sequence (RS); and iii) pseudo-k-tuple reduced amino acid composition. The LeNet5 learning architecture is utilized for the effective prediction of drug-potent proteins. Moreover, the Chi-Square feature selection algorithm is utilized to assess the impact of feature selection on the performance of the model. The proposed approach attained an accuracy of 99.41% on the benchmark dataset using 5-fold Cross-Validation, which indicates that the proposed model performs significantly better than the state-of-the-art approaches. The results obtained in this study indicate that the proposed method can be trusted and utilized by the research community for effective prediction of drug-potent proteins, and consequently, to advance the field of drug design and development
Transgenic Valencia orange and sour orange challenged with Phytophthora: a gene expression analysis
No full textCitrus is an important fruit crop grown globally for its nutritional and commercial significance. Texas ranks third in United States citrus production, with most of the crop being grown in the Lower Rio Grande Valley (LRGV). Phytophthora is a ubiquitous soil-inhabiting
pathogen in the region that causes citrus diseases of economic concern. A plant’s resistance to a pathogen is determined by its ability to elicit an immediate defense response to the infection of that pathogen. There are many defense-related genes in plants shown to confer resistance to specific biotic and abiotic stresses. Cyclic nucleotide gated ion channel (CNGC) plays a role in
fungal and oomycete resistance. Transgenic Ruby Red grapefruit and Valencia sweet orange trees harboring a CNGCcit transgene, and a single sour orange seedling exhibited a high tolerance to infections of P. nicotianae in previous studies. In this study, the transgenic and nontransgenic Valencia orange trees were challenged against Phytophthora and leaf tissue was collected from each tree at 0-, 24-, 48-, and 96-hours post-inoculation, to analyze the expression of genes related to plant fungal and oomycete pathogenesis. Sour orange cuttings propagated from the single P. nicotianae-tolerant sour orange seedling were also inoculated with Phytophthora and root tissue collected over 96 hours was analyzed for the expression of the same orthologues. Analyzing the expression of these genes provides an understanding on how they differentially express in these plants when challenged with Phytophthora
Comparative study of novel techniques for static power leakage reduction in nano-scale VLSI CMOS circuits and their application to mobile devices
No full textOver the last four decades, the electronics industry has seen continuous technological improvement. User satisfaction has increased since devices are now portable with improved performance. The improvement in battery performance has also been a major concern, and
industry experts continue to carry out cutting-edge research to improve battery life. As we all agree that the development of the Metal Oxide Semiconductor Field Effect Transistor (MOSFET) and the Complimentary MOSFET (CMOS) has been significant to the
miniaturization of these devices, however, as we scale to nano region, these devices dissipate significant leakage power even when they are in idle mode. These leakage power losses become a fundamental issue as millions of these transistors are fabricated on a single silicon wafer. This research presents a comparative study of two novel approaches: the Sleepy stack with leakage control transistor (LECTOR) transmission and the Adaptive optimal body-bias voltage method and their application to mobile devices. The design circuit performance is analyzed for 32nm, and 45nm technology in LTSPICE, measuring power losses from 0.5v to 1.5v at 27°C, 50℃ and 100℃
Groundwater modeling : investigating the water levels of the Southern Carrizo-Wilcox Aquifer within the Winter Garden Area
No full textGroundwater is considered one of the main resources that provides high quantities of water for everyday usage. In Texas, especially in the south, the main usage for groundwater will tend to range from irrigation for agriculture, oil, and gas production such as hydrofracturing and potable supply. We obtain this resource by pumping it from an aquifer, which is known to contain large quantities of water. However, due to the passage of different permeable material that the water itself undergoes, the chances that the water will be contaminated with metal ions such as cations and anions are a possibility. In addition to the metal ions, high levels of salt will render the water non-usable for use. Finally, due to high demand for this precious resource and declining water levels because of natural causes such as drought, many aquifers face environmental stress. The lack of water availability is limited by fact that most groundwater is composed of many salts that it is categorized as brackish, and the water that is considered freshwater only makes up a fraction of what the aquifer contains.
The purpose of this study is to examine water availability and water levels for an aquifer known as the Carrizo-Wilcox Aquifer within the Winter Garden Area. The Groundwater Availability Model (GAM) files from the Texas Water Development Board (TWDB) will be used to examine changes in the future by investigating trends of groundwater levels at actual well locations and comparing it to historic well data and trends
Exploring body image among Hispanic dancers
No full textBody image has long been an area of concern for adolescent females. Particularly, dancers who have been known to push body image ideals and standards of beauty. The dance community has been known to encourage a specific slim physique and through a sociocultural
framework, the present study aims to explore ideal body image among Latina adolescent dancers. Participants completed an open-ended survey that explored their ideal body, external influences and pressure. Through rigorous coding, four main themes emerged: Traditional Dance
Body Standards; Body Inclusivity; Perception of Body Type; and External Factors. Themes and sub-themes that emerged show a range of opinion regarding body type. These findings are important for community dance instructors, parents and teenagers alike because they can see there is a general desire for more inclusivity in dance. Limitations and future directions are included
Meta-analysis of the effectiveness of MyMathLab in college algebra.
No full textCollege algebra is regarded as a gatekeeper course for attaining degrees, higher paying jobs, and overall individual empowerment. Educational stakeholders have reconsidered alternative instructional approaches such as computer-assisted instruction (CAI) to curb high failure rates and increase retention rates among college algebra students. One type of CAI, MyMathLab (MML), uses adaptive multimedia features to stimulate students’ interest and create a learning environment that is more engaging. This research studied the effectiveness of MML in college algebra. Understanding the effectiveness of MML will improve the knowledge base necessary to make informed decisions about how to close the achievement gap among college algebra students providing them access to better life opportunities. As such, the purpose of this study was to conduct a meta-analysis to determine the overall effectiveness of MML among college algebra students. The theoretical framework applied in this study was B. F. Skinner’s theory of behaviorism (Operant Conditioning) due to MML’s ability to reinforce mathematical efficiency by modifying behaviors through incentives. The overall effectiveness of MML among college algebra students is the research question that guides the study. A meta-analytic design was chosen so that mean effect sizes could be pooled from different primary studies to infer more accurate generalizations. The primary studies collected from an exhaustive literature search with stringent eligibility criteria served as the sample in the meta-analysis. Hedge's was used to determine effect sizes because of its superiority at addressing biases when sample sizes are small. The data pooled from a total of (= 7) quasi-experimental primary studies that summarized the mean effect sizes using Hedge’s in a forest plot was based on the random-effects model. The results of the meta-analysis yielded a significant effect (= 0.201, = 0.037, 95% CI [0.012 to 0.390]) favoring MML over traditional college algebra instruction with a moderately low amount of heterogeneity in true outcomes (Q = 9.392, = 0.153, 2= 0.020, 2= 31.130%). Based on the meta-analytic findings, supplying educators with computer-assisted software such as MML could help modernize the curriculum and prevent future generations of college algebra students from repeating the same patterns of struggle and missed opportunities
Real time moving and static vehicle detection with UAV visual media
No full textOne of the increasingly important aspects of machine learning is vehicle detection from
unmanned aerial vehicles (UAVs) in the stream of military, commercial, entertainment, and
traffic applications. Many schemas have been proposed so far for detecting vehicles from an
aerial view and those schemas could not attain sufficient accuracy rates for detecting small
vehicles and moving vehicles owing to the low-quality of UAV videos and the movement of the
platform with varying weather conditions. To improve the accuracy in detecting multivariant
types of vehicles that are static (those whose location is tracked and found to stay the same from
one image frame to another) and in motion (those whose location is tracked to be changing from
one image frame to the next), this research uses deep learning algorithms trained on a custom
dataset. The algorithms that are used are fast and high-performance object detection algorithms
such as You Only Look Once (YOLO) v3 and Single Shot Detector (SSD) algorithms that are
used to detect vehicles in visual media formats taken from UAVs. This research proposes a novel
stacked model for moving and static vehicle detection from aerial vehicle images to reduce false
alarm rates and improve the accuracy over existing systems with pre-processing enhancement
and training with custom aerial vehicle image datasets. The proposed stacked model in this work was observed to have an accuracy of 91% and a processing speed of 60 frames per second, which
was more accurate and faster than other state-of-the-art methods
FPGA implementation of a Recurrent Neural Network
No full textArtificial Intelligence (AI) is an important technique that can be used for a wide range of applications to speed up workflow and produce more efficient results. A Recurrent Neural Network (RNN) controller designed for controlling solar energy integration into the bulk power grid in real-time was implemented on an Intel Cyclone V Field Programmable Gate Array (FPGA) board and then compared with its corresponding MATLAB calculations, to solve the implementation difficulty: limited calculating and memory in an embedded environment. FPGAs are reprogrammable digital logic chips, and various electronic design automation tools are available through which these boards can be programmed. In this thesis, Intel Quartus Prime was used to program the Field Programmable Gate Array. The Intel FPGA IP library has specialized mathematical blocks to efficiently perform fixed-point and floating-point computations; in this work, the specialized 32-bit floating-point blocks were used to design the Recurrent Neural Network implementation, which is divided into three layers, with three main challenges: clock latency, design of the activation function, and matrix multiplication. This work addresses different ways to overcome these three challenges to yield optimized results
Parental involvement impact on student achievement in an elementary setting with low-socioeconomic status.
No full textEducators believe involving parents in their child’s education is valuable for success and is crucial for a child to reach their fullest potential. Parental involvement affects students' academic performance from families of low socioeconomic status. To better serve the needs of students from families of low socioeconomic status, it was necessary to analyze parental involvement and its impact on student achievement. Parental involvement can improve a student's academic performance. Still, the literature has not yet identified how parental involvement improves academic performance for students of low socioeconomic status in an elementary setting. To address this gap, a quantitative study was conducted using non-identifiable archived data to investigate the impact of parental involvement on student academic success. The research explored the efficacy of parental involvement factors, such as the effects of lack of parental involvement and inflexible work schedules. The sample studied consisted of students’ non-identifiable archived State of Texas Assessments of Academic Readiness (STAAR) data using a random sampling generator for equal opportunity and four different parental involvement events from grades 4th–6th at an elementary setting for 2018-2019. Based on the data, this study confirmed a positive correlation between low-income parental involvement and student academic achievement using a Pearson r test. The significant positive relationships identified for all four events (event 1, event 2, event 3, and event 4) indicated that as parental involvement increases, so does student academic achievement in grades 4th-6th. The study contributes to the broader understanding of how parental engagement can influence academic achievement while highlighting the relation between student academic achievement and grade levels. As parents continue to be a part of their child’s academic success, it benefits their overall academic achievements. Implications and best practices for analyzing non-identifiable archived data determined the impact of parental involvement and how it correlated with the success of student’s academic achievement are further discussed
RFID mobile robot simulation in a warehouse model using ROS-GAZEBO
No full textRadio Frequency Identification (RFID) has become greatly prevalent in the industry especially for warehouse management which shows ability to improve inventory accuracy, increased operational efficiency, augment supply chain and more. It greatly reduces costs and has advantages to become more prevalent in the industry. RFID systems have the scope for tasks like object identification, localization, information access etc. Establishing the type of RFID system for an application depends on the task and requires a thorough understanding of the factors and knowledge of its attributes affecting the RFID system. The proposed work demonstrates an open-source RFID simulator using ROS that allows modelling of an environment, a mobile robot and testing RFID system. This way, we analyze a localization algorithm and study the various weaknesses that arise attempting to fix before it is deployed on the field. Several tags and antennas can be placed in the world environment, and the simulator runs a probabilistic approach for each tag placed utilizing the RSSI signals. The reader or antenna can be parametrized with different configurations to reproduce a specific requirement. The running simulator is GAZEBO that uses Robotic Operating System (ROS) framework to connect to a larger audience