ABSTRACT. We will welcome you to the ACM Mid-Southeast Conference and introduce our speaker.

ABSTRACT. Neuromorphic computing is a popular technology for the future of computing. Much of the focus in neuromorphic computing research and development has focused on new architectures, devices, and materials, rather than in the software, algorithms, and applications of these systems. In this talk, I will overview the field of neuromorphic from the computer science perspective. I will give an introduction to spiking neural networks, as well as some of the most common algorithms used in the field. Finally, I will discuss the potential for using neuromorphic systems in real-world applications from scientific data analysis to autonomous vehicles.

ABSTRACT. Role playing exercises are a pedagogical teaching technique to engage students and help them understand programming concepts. Four different exercises used in classroom setting are discussed herein. Early ideas such as differentiating functions of processor and memory can be modeled by a person doing a computation using large numbers on the whiteboard. The person models the processor and the whiteboard the memory. Different sections of code can be modeled by a person as well. This is useful for illustrating looping mechanisms, where students can identify which iteration gets access to the whiteboard - modeling the data and the output console. Try-catch blocks can be explained with each block being played different student, to illustrate flow of control through the program. Lastly, understanding the difference between classes and objects can be modeled by students role playing objects, where they get handed note cards with the data they 'manage' and the methods (processing) that is defined on those data. In the context of teaching programming, it is most constructive to indicate before or during the exercise what the students represent. A scripted role play, as opposed to vague role play used in experiential learning, is better suited to tie in the student actions and executable program code. Overall, I have found that explaining the metaphor beforehand or during the role play, and tying it to code written and executed in class is best for maximizing student understanding.

ABSTRACT. In Fall 2021, we transitioned UT Martin’s second-level programming course (CS1.5) to Linux. This move was designed to give students a more gentle introduction to our Linux toolchain, with an aim to improve student confidence in Linux development. GitHub classroom has played a vital role in this transition, providing a simple tool for assignment submission, automated testing, and grade/code collection.

ABSTRACT. The Tabletop Interactive Play System (TIPS) was created to provide linguistic input to deaf and hard of hearing (DHH) children while exposing hearing parents to American Sign Language (ASL) through an ASL vocabulary recommendation system during non-obtrusive toy play. Along with our novel ASL recommendation system, a definition of successful ASL delivery is introduced to assist in determining the quality and quantity of the ASL input for the child. We compare our proposed system with pre-established technology that can aid ASL learning. Through our proposed measurements and future user study, we will be able to learn more about DHH children and how they respond to ASL, as well as objectively determine the best system to aid parents in learning ASL.

ABSTRACT. Voice-based authentication systems are susceptible to being bypassed via machine-induced audio replay attacks. In these attacks, a machine-induced sound is played to the audio-based authentication system in an attempt to gain unauthorized access by impersonating a legitimate user of the system. This research work will apply machine learning to detect audio-based replay attacks by differentiating between audio recordings induced by a live (human) speaker and a digital recording of the same speaker.

To this end, we will collect voice samples from participants and then replay those voice samples from a mechanical device to get two different sets of recordings from the same participant – one of the live speakers and one of the replays. We will convert the raw audio files to spectrograms to implement computer vision and image processing techniques to extract image-based features from these samples. Then, we will use these features (extracted from the spectrogram images) to train and create a machine learning model. After that, we plan to test and report the model's accuracy in determining which images are generated from replay attacks and which are generated by a live speaker. The objective of this research is to assess the capabilities of computer vision and machine learning techniques to detect replay attacks on a voice-based authentication system. Finally, we plan to evaluate the accuracy of these tools and techniques to differentiate between the replay attacks and live speakers.

ABSTRACT. Despite the efforts of government legislature and non-profit organizations, the global rate of human trafficking remains persistent. The challenge stems from a lack of qualitatively structured data. However, technological advances have caused many traffickers to develop hidden online advertisements for sexual services. This presentation discusses the legal efforts in place and then explains the leverage of online trafficking ads through the development of a web crawler to identify these ads. The goal of this research is to aid law enforcement in capturing traffickers through the identification of the ads as well aid non-profit organizations in connecting with survivors who seek help.

ABSTRACT. Text summarization is a massive research area in natural language processing. It reduces the larger text and provides the prime meaning of a text document. Finding the meaning of the larger text is needed for proper text analysis, which a better-designed summarizer is entitled to do. Abstractive summaries emulate more closely how humans summarize. We put together new sentences that aim to tell the whole story, but from a higher-level perspective. However, abstractive summaries are technically difficult to produce. Considerable work has already been done in Abstractive Text summarization of high-resource languages like English. When it comes to a low-resource language like Bengali, because of its syntactical complexity and morphological deviation, it’s not easy to produce an efficient Bengali text summarization methodology. In recent years, a few researchers have come up with new ideas and applied existing summarization techniques in abstractive Bengali text summarization. One of the key limitations of existing solutions is that they are primarily concerned about extracting facts from the source text but overlook other crucial factual information, such as the related time, locations, reasons, consequences, purposes, participants, and involved parties. Furthermore, the current summarization frameworks are inadequate in modelling the complex semantic relations between facts and the corresponding information, leaving much room for improvement. This study uses FFSUM model in the Bengali dataset for the first time so that the facts and facets of the data can be explored and increase the efficiency of the state of the art of Bengali abstractive text summarization.

ABSTRACT. In today’s highly competitive business world, it is important for businesses to predict the turnover rate (often called “Churn”) of their customers as accurately as possible in order to devise effective and efficient marketing and customer retention strategies. The Churn prediction is considered a classification problem in the field of data sciences. The main goal of this research was to develop machine learning models and compare their performances for a typical Churn problem encountered in a banking environment. The performances of these models were compared to each other in order to identify the best models for the churn problem. The machine learning models used in this research included 15 different artificial neural networks (ANNs), logistic regression, and random forest. The data was obtained from Kaggle website and contained 10,000 records and each record included 10 customer features and a binary target variable reflecting if a customer stays or leaves the bank. The data features included both numeric and categorical values. Python language on Google COLAB platform was used to develop the models. All models were trained with 80% (8000 records) of total data and tested with the remaining 20%. The results of these models indicated that all ANN models had very similar performance with an average prediction accuracy of approximately 86% ± 1%. A deep learning neural network with triple hidden layers and a total of 14 hidden nodes, and a random forest model with 101 trees provided the best prediction accuracy (~87%). However, the training time (of under a minute) for the random forest model was significantly lower than ~25 minutes training time (corresponding to 100 epochs) for the neural network model. Therefore, it is recommended that random forest models be the top choice for classification problems like the one described in this research.

Mentor: Dr. Masoud Naghedolfeizi

ABSTRACT. Navigation optimization analysis: Comparing algorithms for the Travelling Salesman Problem Abstract Himanshu Bohra, Donghyeon Park

This study shows a comparative analysis of the multitude of algorithms used to solve the ‘Travelling Salesman Problem’ such as the 2-opt algorithm, and strategies such as dynamic programming, with multiple scenarios to explore hidden costs and benefits. The goal of the study is to find out a general solution for determining the at-hand position in the cost-benefit graph of these chosen algorithms, and thusly find out which algorithm would be best to use in real-time. This study can be primarily applied to robotics navigation, or any other real time navigation where the given data-set may change abruptly and unpredictably.

ABSTRACT. In the field of information sharing there is a need to pass information among users in a robust, fast, and secure manner. This project deploys a Telegram bot to communicate with an SQL database utilizing a secure token. This secure to- ken allows the identity of the recipient’s to remain anonymous and secure. This approach allows for many possible applications such as a manger reaching out to many employees within a department, a data server reporting to administrators when parts of the system are down, and even a piece of code alerting a user that the code is completed or passed a required threshold. The SQL database is utilize to keep track of the groups that need to be addressed allowing the user to focus on the message being sent and the database to do the tedious work of finding the users in a given group. Telegram was chosen as it is widely used, known for being secure and providing detailed documentation. The goal of this project is to develop a dynamic API that can be utilized and shared by others.

ABSTRACT. Modern operating systems students seem to be uncomfortable with C programming when they enter the class. This means that they are trying to gain familiarity with C and low-level programming at the same time. However, time is at a premium and a lengthy review of C would take too long. That is where the assignment in this talk comes in. By using an older Linux distributed in a simulated environment, students are invited to carry out the buffer overflow attacks described in the classic article "Smashing the Stack for Fun and Profit" by Aleph One. In a possibly irresponsible twist, after successfully attacking the code output of late 1990s compilers, the students are then asked to analyze and attack modern gcc output and attempt to evade the stack protectors. This assignment has the appeal inherent to forbidden knowledge and just enough high and low level concepts of C to serve as a worthy review to prepare students to manipulate memory and machine code. This talk will detail this assignment, its software setup, and will end with a discussion of student outcomes.

ABSTRACT. Hackers usually use compromised hosts to launch cyber-attacks in order to reduce the chance of being detected. These compromised hosts are called stepping stones. With stepping-stone intrusion (SSI), an attacker uses remote-login tools such as SSH to access a chain of stepping-stones and send attack packets to a remote target through the intermediate stepping stones. The purpose of SSI detection (SSID) is to determine if a computer is used as a stepping-stone. In this paper, we develop an efficient detection algorithm for SSID to determine if a computer is used as a stepping-stone host by calculating and comparing packet crossover ratios. Packet crossover ratios can be easily computed, and thus the detection method we proposed for SSID is not only efficient but also can be easily implemented. Properly designed network experiments are conducted to verify the effectiveness and correctness of our proposed innovative method for SSID.

ABSTRACT. We want to test the bounds of citizen science; citizen science is the practice of public participation and collaboration in scientific research to increase scientific knowledge. On April 8th, 2024, spanning from Texas to Maine, we seek to involve millions of citizens in an “Eclipse Megamovie”. In 2017, a similar project was produced with over 1000 citizen scientists, and we seek to do this on a significantly larger scale. As a team, we are developing a user-friendly application for smartphones that will utilize camera tech to take a series of photos and videos of the Eclipse, all of which will be sent (along with the timestamp and precise location) to NASA for analysis and compilation. The app is planned to be tested during the Southern hemisphere hybrid eclipse of April 20th, 2023, and the US annular eclipse of October 14th, 2023, and released in advance of the April 8th, 2024 Eclipse.

ABSTRACT. Teaching upper-division computer networking related courses for undergraduate in a distance learning environment possesses a number of challenges in both technical and pedagogical aspects. A hybrid approach that utilizes a remote networking laboratory, virtual machines, and specialized simulation software makes it possible to offer an IT degree with a concentration in Computer Networking possible. In addition to offering courses in the format of DVC (Desktop Video Conferencing, in which a course is taught live with video-conferencing), the virtualization technology plays a key role to simulate complex network infrastructures and server technologies. We will explore and evaluate different strategies to set up the virtualized lab and accommodate essential practical skills for several network courses. Based on the student feedbacks and grades, student learning experience is even better compared to a traditional lab in which it may not be possible to deploy a complex network. Students are prepared with the knowledge and skills to take those in-demand IT certifications. Future plan will be discussed about several academic programs from leading IT companies of providing free e-learning course shells and online lab.

ABSTRACT. Enterprise content management (ECM) cloud computing is a cutting-edge technology concept that enables a collaborative approach to information while assisting businesses in managing, storing, and organizing their information effectively. With ECM Cloud, businesses may significantly streamline business processes, improving the efficiency of daily operations. The first step in developing work from anywhere environment is connecting people to the content they need, when and when they need it. More than 75 percent of businesses now use a hybrid model that predominantly employs remote workers, necessitating a bigger than ever requirement for effective operations and communication. This research aims to focus on the Content Cloud that enables modern work by giving clients the means to collaborate and exchange information across business systems, companies, and locations to connect their workforce. Businesses enhance productivity by providing workers with solid solutions that are both secure and easily available. By seamlessly integrating material across crucial business systems like CRM, ERP, and HRIS, it will be possible to make sure that the right content is available at the right time and in the right context. Automating content-centric business activities in the cloud increase efficiency and satisfy shifting consumer and industry demands. For enterprise content management, the Content Cloud is a suite of full life-cycle management and archiving technologies. Through the connection of Content Cloud with information production and consumption systems, enterprise-grade content management is extended further throughout the organization. The goal of this study is to use text mining, OCR data classification, and machine learning techniques to examine the distribution and ease of access

ABSTRACT. Elementary school is a critical learning period for students and developing early math skills has been shown to lead to higher success than any other elementary subject. During the elementary years, children will gather important mathematical logic skills and strengthen the ones they have already learned. Children are building the fundamental tools they will use for overall math comprehension and real-world situations. This research uses Hoppers Fables, which was created to extend those math skills in a captivating manner for children to gain basic computing skills. Beyond teaching skills, it is important to push children towards interests of math and develop a healthy relationship with the subject. In Hopper's Fables, Blockly, Google's drag and drop programming language, is used to develop an easy and comfortable format for the age group for which the website is created. Hopper's Fables creates a smooth storyline of a character and their basic day to day activities at school. This presentation reviews the Hoppers Fables creation and application, specifically highlighting new additions implementing a deeper story plot and more mathematical logic in addition to pilot study data.

ABSTRACT. A great deal of scientific progress has been made in the world of prosthetic limbs. We have prosthetic limbs that can pick up signals directly from the brain and perform desired functions as a natural limb would do. But with these breakthroughs, the cost of such devices has also risen dramatically. The Modular Prosthetic Limb (MPL) developed by John Hopkins Applied Physics Lab (JHAPL) costs around $500,000. While there are cheaper options on the market, most prosthetic limbs are too costly for a middle-class family to purchase and maintain. The goal of this research is to determine if it is possible to create an affordable but effective prosthetic limb prototype using a six degree of freedom robotic limb and a motion tracking device. This presentation provides an overview of existing prosthetic limbs along with the results of an initial evaluation of the robotic limb with the motion tracking device.

ABSTRACT. With increasing cyber-attacks in today's data-driven world, the demand for the database security professional is increasing. Teaching database security in an undergraduate database administration course serving both computer science and cyber security students can be challenging. This work discusses different techniques to secure the database engine and the stored data in it and also identifies relevant hands-on demonstrations to help students better understand the concepts and gain real-world insights into database security and auditing. Different techniques to ensure database security using data encryption, authentication, authorization, and change tracking are discussed.

ABSTRACT. Web Graphics Library (WebGL) is a JavaScript API used to render 3D and 2D graphics in web browsers. Getting started with a WebGL program can be an overwhelming task. This is because there are many parts that need to interoperate to render a graphics scene. From a software architecture perspective, we can use a monolithic approach but doing so makes the program difficult to maintain and reuse. I will present a software architecture that promotes maintainability and reusability. Maintainability is achieved with abstractions where each abstraction represents a major part of the architecture. Reusability is achieved by creating a model to represent the architecture and providing a templated implementation that can be easily cloned from GitHub as a starting point. Improved testability is an additional benefit of using abstractions. I will also present how to use the QUnit testing framework to test the various implementations of the abstractions.

ABSTRACT. During and after the Covid-19 pandemic the IoT devices has been widely deployed in homes and offices where the devices can be compromised by the physical and cyber-attacks. Mutual Authentication is an authentication process that verifies each device. In this study, we show mutual authentication protocol between IoT devices using hardware security module, called Zymkey. Zymkey is a small hardware security module that can work with Raspberry Pi and provides useful security functionalities such as multifactor device identity and authentication, data encryption and signing engine, key generation and secure storage, and physical tamper detection. This study shows how to set up mutual authentication protocol between IoT devices and how to prevent the data stored in the IoT devices from being illegitimately retrieved or stolen using Zymkey. For the physical security protection, this study also shows physical perimeter and tap detection methods that can send an alert message using email to the system manager. This study can provide some additional physical security to the IoT devices.

ABSTRACT. The time it takes for a child to learn how to drive a motorized wheelchair is equivalent to a teenager learning how to drive a car, which is around 40 to 50 hours. Motorized wheelchairs can cost anywhere from $2,000 to $15,000, which insurance companies will help pay if the child can demonstrate knowledge of how to use the chair. Children can learn how to drive a chair by using a rental, which costs additional money, or using a temporary loaner; however, the learning process takes time. This presentation introduces Driving the Mountains, a car game allowing children to improve their physical and cognitive ability using a joystick without the need for supervision. Moreover, the experience of using a joystick enables Physical Therapy staff to learn the best joystick placement for the child. Driving the Mountains is intended to be fun while teaching the child how to use a joystick and control a vehicle.

ABSTRACT. In recent years, higher education institutions have been compelled to offer relevant academic programs in information security to meet the growing demand for trained cybersecurity workforce needed in both the public and private sectors. The field of cybersecurity is relatively new at higher education institutions. Consequently, many educational institutions lack the infrastructure and expertise necessary to quickly establish a program in this area. This problem is even more acute for smaller colleges and universities, which typically have very few resources at their disposal. This paper describes the challenges that small colleges and universities face when attempting to acquire the resources they need to build an effective cybersecurity program. The paper also discusses the various incremental steps that these institutions might take to eventually develop extensive curriculum in this field.

ABSTRACT. This project involves ongoing efforts to create an environment for physical robotics and computer vision experiments. Our primary goal is to provide tools and procedures to help students in undergraduate research within NASA grants. Our research environment includes TurtleBot robots integrated with LIDAR sensors, NVIDIA Jetson Nano edge computers, and the ZED color and depth camera. Students in the lab carry out the system integration tasks, followed by installing the Robot Operating System (ROS) on the NVIDIA Jetson Nano edge computers. Advanced software library modules for image classification, object detection and machine learning can be used individually or integrated to create advanced and hybrid computational models used in robot navigation and mapping applications. The initial direction of the undergraduate research supported by this environment is related to the swarm of robots movement for collaborative scene understanding and object recognition. Our student research environment generally supports many other critical aspects of robotics and computer vision. These areas are undergoing dramatic changes with the development of new Artificial Intelligence and Machine Learning algorithms. Due to these rapid changes and the advanced nature of the work, it is challenging to include recent developments in undergraduate research in a meaningful way. Therefore, the scaffolding system has been developed as an extensive support environment consisting of programming templates. The software templates are provided to students to scaffold their learning, proceed with research experiments in a timely fashion, and avoid unnecessary impediments like developing numerous software components from scratch. *This work was partially supported by the NASA MUREP Space Technology Artemis Research (M-STAR) Grant.

ABSTRACT. Our project is a 3D real time game in the Unreal engine. The game is a fantasy genre in a medieval setting. The game will have 3 levels in it. It will have a variety of enemies and weapons. There will be magic. The different types of magic for the player to use would include pyromancy, sorcery, and faith based magic. You can upgrade weapons and get access to abilities in the game. The upgrade system, and the ability system are meant to give the player the ability to make a build that fits with their play style. With each level the enemies will gain hp and do more damage. Each level will have a boss enemy in it.

ABSTRACT. In recent years, there has been an increasing demand for reliable automatic license plate recognition technology in both private and governmental organizations. Based on recent innovations in machine learning and image processing, it was hypothesized that it would be possible to develop a smart detection technology to identify vehicle license plates under normal and adverse environmental conditions with relatively good accuracy. To investigate this hypothesis, we developed a license plate recognition model (LPRM) to examine the reliability of this technology under the aforementioned conditions. The Python language was used to develop codes for the LPRM. Several libraries such as TensorFlow, NumPy, CuPy, Matplotlib, and OpenCV in conjunction with a previously developed object detection model were employed in the model development. An Optical Character Recognition (OCR) technology, EasyOCR, was integrated into the model in order to extract the characters from the license plate. The license plate recognition aspect of the model involves resizing the image to a width of 600 pixels while maintaining the image’s aspect ratio. The resized image is further processed to create detection boxes around all the detected objects within the image. Finally, the detected objects are filtered in order to find the license plate specifically. The accuracy of the LPRM was examined by using 22 different images of vehicle license plates under normal imaging conditions. The model was able to detect the license plate characters at 75.52% accuracy. The adverse environmental conditions were simulated by distorting selected images with a Gaussian Blur on the X and Y-axes. The model was able to detect the license plate characters with an accuracy of 67%. In conclusion, we found that our license plate model was reasonably accurate. Although our OCR engine was not as advanced as many others, we could still produce relatively accurate results, even under non-ideal conditions.

ABSTRACT. Sentiment analysis and recommendation systems are among the most active areas of research in machine learning. Sentiment analysis focuses on using natural language processing and text mining techniques to evaluate the sentiment (i.e. positive, negative, neutral) of an unstructured text such as a tweet, comment, or product review. Recommendation systems use a variety of data science techniques to generate personalized content recommendations for the users. Here, we present a Python-based prototype for recommending songs to the users based on the sentiment of their reviews. We used the Amazon reviews and the Spotify music datasets from Kaggle for development purposes.

ABSTRACT. The affordability and accessibility of open educational resources (OER) have made them a rising star in the education era. Apart from cost competency, there is a growing body of evidence suggesting the efficacy of OER adoption and use in relationship with improved learning performance and retention indices. However, in contrast to the reported incentives, barriers exist to the successful and widespread adoption of such resources, including a lack of knowledge and confidence in quality. The current talk discusses locating, evaluating, and selecting OER resources and their impacts.

ABSTRACT. Our project is Project Spellda, a two-dimensional, top down role playing game developed in Godot. Our game will minimally include 3 different levels: a tutorial level to introduce the mechanics of the game, followed by an overworld and dungeon that are intertwined. At the beginning of the game, the player will choose two of the four in-game elements: Fire, Water, Air, and Earth. The player then explores the surrounding area, looking for upgrades to their default spells. At any time, the player can dip into the dungeon to test their abilities on the stronger enemies that await them, with the end goal being to get to the center of the dungeon and slay the boss. The game is made entirely using Godot's built-in scripting language GDScript.

ABSTRACT. Magnetoencephalography (MEG) is the measurement of the magnetic fields generated through neural impulses produced by the brain. A MEG signal contains important information about the health of the brain. By analyzing these magnetic fields, it is possible for it to detect various neurological diseases. Denoising the signal will make it much simpler and more effective to analyze the data and diagnose a patient. We have used two alternative methods in this research to analyze the MEG data. One method for processing and aiding in data visualization is MNE-python. TensorFlow is then utilized to build the model and train it until it reaches an acceptable level of accuracy. Data in FIF format will be sent to the model for training, validation, and testing. The model will be based on supervised learning and deep learning.

ABSTRACT. I have long been interested in how to improve the outcomes in the introductory programming class. These courses often have a failure rate of over 50%. One suggestion has been to switch the initial language students learn from C++ to Python. Anecdotal evidence from my teaching Python first over the last ten years certainly shows a significant improvement in that metric. Questions remain, however.

The first question is whether in making the initial programming course easier to pass, we are just deferring the inevitable by one semester. If that 50% of initial programming students are going to fail their second programming class, have we done them any favors? The second question is whether the version of Python taught in introductory classes leads to the idea of Python as a beginner language only, and might stop further student development in it. A related third question is if the beginner Python we tend to teach is so far removed from how professionals use Python that we are creating bad habits that will have to be unlearned later.

In particular, professional-level Python tends to focus on the functional features of the language. Much of Python’s success in fields such as Artificial Intelligence and Data Science is a direct result of these functional features. Yet, we often teach Python as “C++ light” and spend a great deal of time on imperative features of Python that are not consistent with the functional aspects of the language.

In my presentation, I plan to focus on the above questions. I will conclude with some ideas on how we can integrate more functional Python into our programming curriculum.

ABSTRACT. Kronos is a virtual reality video game with roguelike elements in which time is falling apart at the seams. It falls on the protagonist, Kronos, to restore the timeline to its chronological order. Along the way, the hero is met with obstacles to overcome to achieve this goal. Kronos has pixelated 2D art sprites and textures in a 3D world that is inspired by games like DOOM (1993) and Paranautical Activity. The game is a first-person shooter where you are traveling between a mismatch of time periods in order to restore the original order of time. Kronos is built using Unreal Engine 5 and a Procedural Dungeon Plugin created by BenPyton. The behaviors of the enemies within the game, the item effects, and the weapon mechanics are created using Unreal Engine’s Blueprint system, which is a visual scripting language. Each floor is set in a different time period with various elements about each period feeling out of place. This can be seen in the enemies and objects within the level.

ABSTRACT. Magic Mirror

William Carroll, Juwan Hollingsworth, Suraj Patel Department of Computer Science and Information Technology Clayton State University Morrow, GA 30260 {wcarroll, jhollingsworth5, spatel53}@student.clayton.edu

Abstract

In this project we are turning an old non-touchscreen flatscreen tv into a smart mirror that is touchscreen. This smart mirror also known as “Magic Mirror” will be connected to a Raspberry Pi 4. This smart mirror will have the ability to act as a full functional notice board on campus and display events for the organization. For the increase in interaction, there will be visual interaction between users with the included ability of a touchscreen functionality. We are integrating this smart mirror with a camera, Raspberry Pi, 2-way mirror film, RF frame, and a LED monitor. This magic mirror can be placed and designed for any location as a wall mirror mount, table mount, or portable on wheels. This also boosts technology in any establishment. This research will open new doors in innovative thinking among students to help new creation of smart devices such as this, smart mirror. It will have contribution to explore more details to add additional applications for smart mirrors to improve student learning.

ABSTRACT. Computer science drives innovation! Indeed, one need only look at today’s headlines, listen to an economic podcast or review the stock market prices to see the effect computer science and technology has on the world’s economies. With this effect, isn’t there an educational and moral obligation to provide all students with the knowledge needed for a world where computing is ubiquitous? Computers are contained, used and applied to just about every aspect of our daily lives; however, knowledge about computers is not keeping pace. Unfortunately, the U.S. education system has fallen woefully behind in preparing students with the fundamental computer science knowledge, skills, insights, and perspectives they need for future success. While all 50 states allow computer science to count towards a graduation requirement, only 23 require a computer science course. Because computer science is neither explicitly nor discretely part of the “core” courses within STEM, computer science is uniquely isolated, surrounded by a lack of clarity within each state’s requirements. Defining what constitutes a computer science course as well as who has the credentials to teach it are quite divergent depending upon the state one wishes to teach. No two states align with a definition of a computer science nor even what discipline the course should be contained -- a career and technical, business, math or technology course? No area of professional practice generates more frustration among current and aspiring computer science teachers than computer science teacher certification/licensing state pathways. The presentation explores the systemic “bugs” in accessibility to computer science classes in U.S. K-12 schools as well as an analysis for the licensing pathways to address the question: If computer science is the pathway to computational thinking and innovation, then why the disconnect in definition, policy, and opportunity for licensing?

ABSTRACT. A voice assistant is a device that uses natural language processing to interpret user commands and provide some service based on the input. Well-known examples of voice assistants are Amazon Alexa, Google Home, and Apple’s Siri. All of those voice assistants generally work in three stages: speech to text, text to intent, and intent to action. This project aims to improve step two of this process, text to intent, also known as semantic matching. In semantic matching, user input is analyzed and parsed into different values as the program attempts to determine what the user wants from the voice assistant. Semantic matching often struggles to handle complex and context-dependent user input. This lack of generality and complex understanding means voice assistants are best suited to short queries with direct factual answers. This project intends to implement different methods to increase voice assistants capability to handle structurally diverse long user queries by classifying input sequences into intent and keyword pairs to better quantify and explore search spaces for complex queries. This project also aims to implement program directives that allow the voice assistant to use previous queries to contextualize subsequent queries and filter out more information, resulting in an assistant that is ultimately more capable of handling both more complex user queries and less formatted, more abstract user requests.

ABSTRACT. The Emergency Alert System is a system that is developed to warn the general public about Emergencies that threaten the general public, and national security.In 1997 the Emergency Alert System and the IPAWS (Integrated Public Waring Alert System) were designed to be integrated to do 2 tasks, Allow the President to speak on all Stations and Warn the public on more than just National matters but Local matters: like weather warnings. This system in its modern form is convenient, but it’s also flawed in every right due to not adapting to the times.

The Emergency Alert System works in this form. There is a Primary Entry Point System, that will send a warning to Radio Stations and Broadcast Stations, they use a device called the ENDEC (Encoder/Decoder) to decode these alerts. From there it is the station's responsibility to air the alert. Government agencies like the FCC (Federal Communications Division) and FEMA (Federal Emergency Management Agency) are responsible for the publishing of Emergency Alerts. But the Flaw resides in the Stations, Lack of Security, and flawed firmware in Sage Electronics (ENDEC) Systems. According to Ken Pyle, "In short, the vulnerability is public knowledge and will be demonstrated to a large audience in the coming weeks." This is combined with reports of Stations using the default password. This can cause the Web Server that runs the system, to be exploited.

How do we fix it, simply updating firmware on Monroe Electronics/SAGE and changing default passwords, this is a very fixable situation. If the situation is not fixed, false alerts can be triggered, and/or real alerts can be ignored which in an actual emergency can cause a multitude of issues (e.x. Lack of Public Warning)

ABSTRACT. Recent advances of video analytics systems integrated with IoT allow the development of on-board data processing and control systems. It is still challenging to set up the embedded visions in resource-limited devices. For object detection with Raspberry Pi, we used OpenCV and TensorFlow that could provide the solutions for the embedded visions. We decided to dig deeper into Tensorflow specifically the lighter version of Tensorflow called Tensorflow Lite or tflite for short. Once we got Tensorflow lite installed we started with a simple detection program without the bounding box called Image classification using the programming language Python. Once we got the Image Classification to work, we moved on to the next objective which was to add a bounding box to the detection and recorded the results. We used the first model of Tensorflow Lite called EfficientDetLite0 which has an average Precision of 25.69%. Tensorflow Lite has 4 models this being the lowest precision and the highest model EfficientDet-Lite4 being 41.99%. Even using the version with the least precision, we found the model to be fairly accurate and it was able to detect fruits like apples and devices like keyboards and monitors.

ABSTRACT. South Central Kentucky (SoKY) is a region consisting of ten counties: Allen, Barren, Butler, Edmonson, Hart, Logan, Metcalfe, Monroe, Simpson, and Warren. Around 299,945 individuals live in SoKY, and with approximately 16 museums in the region, an estimated 66,694 individuals are left without a museum or historical site in their respective counties. To combat this phenomenon, the Western Kentucky University Extended Reality Lab demonstrates that it may be possible to bring historical and cultural experiences to SoKY, introduce innovative technology to rural communities, and promote tech optimism.

Utilizing interdisciplinary skill sets and creative problem-solving, an interactive virtual museum experience was developed targeting SoKY-specific culture and history. Featured exhibits include Jack Dappa Juke (an interactive blues theatre experience through the Jack Dappa Blues Foundation); WKU & the Civil War (a virtual comparison between our modern campus and the fort that once resided there); and Hymnals of Jonesville (a reflective auditory experience that celebrates a local African American community and the resulting diaspora after WKU’s expansion in the 1960s). Artificial intelligence (AI) was used for visual concepting, original design and artwork, voice narration, and other components presented in this project. UX research and strategy, and personal accounts from SoKY-based individuals, also informed the curation of an extended reality (VR/AR) museum and platform (website and phone) that best fits regional needs.

The primary impact of this project serves to make education and innovative tech processes more accessible in rural communities. By exhibiting how SoKY locals can relate to and utilize innovative technology, there is a direct positive impact on tech optimism or the common belief that technology is beneficial rather than harmful. An increase in tech optimism correlates with an influx of tech careers in the region, keeping more creatives and innovators in rural areas in SoKY and beyond.

ABSTRACT. Gardening Monitor Abstract Himanshu Bohra, John Morales, Donghyeon Park

Components required: Raspberry Pi, Arduino board, ArduCam sensor (camera), DHT11 sensor. Roles of each component: DHT11 sensor provides humidity and temperature data to the Arduino board, ArduCam provides video data, and the Raspberry Pi works as a server to transmit this data to requesting client devices.

Purpose: The ArduCam sensor combined with DHT11 sensor can have versatile capabilities for surveillance purposes. The primary purpose in mind for this project is as a gardening utility: monitoring plants and the environmental temperature they are in. Providing real time feed to gardeners can help them manage their plants better. The system will use a client-server model, where the Raspberry Pi will act as a server (middle-man) connecting the sensors to the client (typically a cell-phone device, Android OS) and allowing data exchange between them.

Team members and workload: Himanshu Bohra will work on the Android Application, John Morales will work on the Raspberry Pi server app, and Donghyeon Park will work on the Arduino programming to send data to the Raspberry Pi server.

ABSTRACT. With the increase in the application of machine learning (ML) in different domains, incentives for deceiving these models have reached more than ever. As data is the core backbone of ML algorithms, attackers shifted their interest toward polluting the training data itself. The credibility of data is at even higher risk, with the rise of state-of-art research topics like open design principles, federated learning, and crowd-sourcing. Since the machine learning model depends on different stakeholders for obtaining data, there are no existing reliable automated mechanisms to verify the veracity of data from each source. Malware detection is an arduous task due to its malicious nature with the addition of metamorphic and polymorphic ability in the evolving samples. ML has proven to be a solution to the zero-day malware detection problem, which is unresolved by traditional signature-based approaches. The poisoning of malware training data can allow the malware files to go undetected by the ML-based malware detectors, helping the attackers to fulfill their malicious goals. A feasibility analysis of the data poisoning threat in the malware detection domain is still lacking. Our work will focus on two major sections: training ML-based malware detectors and poisoning the training data using the label-poisoning approach. We will analyze the robustness of different machine learning models against data poisoning, with varying volumes of poisoning data.

ABSTRACT. The purpose of this project is to investigate the use of 3D modeling and scanning in the construction of lifelike virtual reality tour experiences. Many virtual reality tours use 360-degree camera imagery because of the relative ease of tour creation it offers, but these tours do not allow the user to interact with the environment and explore freely. They are also limited to sites where 360-degree imagery can be employed, which does not allow for the reconstruction of past sites where no imagery exists. With this project, we aim to produce a realistic and interactive tour of some of Western Kentucky University’s oldest buildings, both in their present state and at the time of their construction. To achieve this, the tour is developed in Unity 3D game engine. Buildings are created using 3D modeling techniques with close attention to detail, while more complex or organic structures around campus are moved into the virtual space via 3D scanning and mesh manipulation using modeling software. Accurate terrain is generated using LIDAR data and modified to include more detailed features not included in the heightmap. This work serves as a proof of concept for larger scale projects involving true-to-life interactive virtual reality environments, especially those concerning the reconstruction of past sites. These are an important development because of their potential for use in educational settings and in the community recognition of past events.

ABSTRACT. The purpose of the Unmanned Aerial Vehicle (UAV) Drone Team is to map the inside of caves and implement the data into virtual reality to eliminate the need to physically enter the caves. The virtual reality caves could be used for education or research purposes.

The drone will collect point cloud data from its surroundings using a LIDAR sensor, which will be stored in a Raspberry Pi. The simultaneous localization and mapping (SLAM) algorithm will be used to keep track of drone’s location. Once the LIDAR data is created, it is uploaded to a database. The LIDAR data will be processed and converted for image processing and LAS point cloud creation that will be used to create the mesh implemented in the virtual reality program. That mesh will be added with more detailed texture and image. The image will then be implemented into virtual reality to recreate the cave environment.

The expected result of the project is to create a virtual environment that accurately represents the geographical data collected by the drone. We hope to learn applications of LIDAR data and the use of Unity in creating programs for virtual reality. With this knowledge we hope that cave exploring can be used for easy access for researching and educational purposes without physical injury.

The research and deliverables of the project will be displayed and presented at the WKU (Western Kentucky University) Student Research Conference. An oral presentation and a poster presentation will be set up at this conference. At the conference, the two presentations will focus on two distinct parts of the project, one for the hardware side of the project, and one for the software side.

ABSTRACT. Many different plants require different environmental conditions to cultivate. A green house is an enclosed environment used to grow plants under regulated conditions. The conditions that need to be monitored are temperatures, humidity level, air, sunlight and nutrients. Maintaining proper humidity levels and temperature is an important factor in a greenhouse therefore finding the right model of a humidity sensor will be important. Labview is a software tool that can be used to program, view and monitor the sensors within the greenhouse.

ABSTRACT. This talk will discuss the latest updates to CompileIt. CompileIt is the online IDE, submission and grading system that I have been developing for the last 8 years. We will talk about how fast the latest versions of JavaScript, TypeScript, Node, Angular, FireBase and the various support libraries that CompileIt is built with are evolving. We will talk about how difficult it is (for a one man shop) to keep up with the most current versions. We will talk about what had to be changed about CompileIt and what could remain the same. We will also look at the new “auto-grader with rubric” that I created this summer that the students and I can run. Like most auto-graders, it can check a program’s output for partial and exact matches. But, unlike most auto-graders, it can also check c++ source code for required formatting, comments, strings, keywords, functions and more. It will also assign a grade and show the rubric so that the student can see what they need to improve about their program.

ABSTRACT. In this paper we propose a drone swarm altitude selection algorithm intended to increase flight efficiency along routes with unknown wind conditions. The algorithm evaluates favorability of present wind conditions to calculate the best altitude to navigate. The energy consumption model used during simulation considers the changes in airspeed required by each drone to maintain position relative to the swarm in various head and crosswind conditions. The algorithm is designed to leverage a swarm’s ability to distribute work to sample available altitudes in search of favorable wind conditions. Performance metrics were generated from simulating a swarm utilizing the flight behavior being tested as well as groups of swarms that flew along the same route at a constant altitude; the performance of the two swarm groups were then compared. It was found that the approach of changing the swarm’s target elevation based on in-flight acquired wind data noticeably improved the swarm’s flight efficiency. During simulation, it was observed that swarms using the test behavior outperformed swarms flying at a fixed altitude across almost every tested flight route.

ABSTRACT. Farming has been around since the beginning of civilization and while the concepts of farming haven't changed there are more threats now than preceding generations of farmers faced. With the rise of automation and remote data, gathering comes the threat of cyber attacks on equipment and data, and that can have devastating effects on the supply chain and in turn the economic growth in the United States. There are many examples of technology creeping into rural farming. Tractors are equipped with GPS units to track plowing and harvest. Chemical sprayers have sensors to determine when and how much insecticide to apply to given crops. Cows have sensors to record how much they eat, and how much milk they produce. Each of these could have dire consequences if the data is not handled properly. In my presentation, I intend to assess some of the threats facing agriculture and offer possible solutions to help prevent attacks from occurring.

ABSTRACT. The subject of class and its object is a fundamental concept in OOP languages. For a class, it is known as a new data type, specifically, a new reference data type. With the class, an object could be declared that is known as the class instance, or commonly known as the object. An object means two components: the object notion and object substance. For some computer languages, like Java, the two processes are asynchronized, meaning that the creation of the object substance is a separate process. For other languages like Python, object notion and object substance are performed in the same time. We use a diagram to illustrate the object notion and object substance, that is, the object notion points to the object substance. In other words, the object notion is a pointer. Unfortunately, in most of our textbooks, the distinguishing of the object notion and object substance is ambiguous. In this talk, we will use linked lists to demonstrate more details of the topics.

ABSTRACT. During the past several years, many universities and other organizations have changed their mode of operation in response to the pandemic. Many universities moved from in-person classes to emergency online instruction. However, students and faculty often had little input into the technology utilized. As the pandemic recedes, studies indicate that many students and faculty would like to continue utilizing technology as part of their instruction. This research explores student acceptance of and interest in continuing to use online learning tools by utilizing the UTAUT for m-Learning Model (Huan et al, 2015).

ABSTRACT. As technology advances, including the usage of the Internet of Things (IoT) devices, organizations capture a great value from optimization and automation. At the same time, organizations are also becoming more and more exposed to cyber risks. Cyber risks are prominent across companies in all industries, including the U.S. Department of Defense (DoD) mission systems, systems used by the governments, and not just technologically advanced companies. This is a global threat especially with companies adding data centers across different countries, further increasing cyber risk. There is a need for companies and organizations to have a cybersecurity risk management process in place and also increase budget to invest in cyber security.

ABSTRACT. Several universities have been adapting their curriculum to provide students with a more active hands-on learning experience that will provide them critical knowledge and skills to be successful in their careers. In a capstone class offered by Athens State University, this is their goal, to form teams of students to work on real-world projects during the semester. In Spring 2022, a team of two students was created to work on a research and application project that builds off of a previous group's work from 2019 that allows for the selection and composition of cyberattack models. The focus of the current team was to add functionality to an application that stores cyberattack models, so that these models can be selected and composed in parallel to create a system attack model representative of two cyberattacks taking place at the same time. In its previous state, the application allowed users to view, filter, import, edit, and select models for sequential composition. The Athens State University capstone team developed a method to auto-generate the parallel composition of two cyberattack models selected by the user. This paper provides insights into the underlying motivations of this project and the experiences gained by the capstone team.

ABSTRACT. In Alabama there are two choices for middle school computer science classes, CS Discoveries and CS Makers. In this paper we discuss the content and implementation of CS Makers as an alternative choice to CS Discoveries. The curriculum, using micro:bit hardware, is based on our NSF INCLUDES work and is supported by the Alabama State Department of Education.

ABSTRACT. Abstract. In this work we cast the automatic reading of dials, gauges, and instruments as a Machine Learning (ML) problem. It is useful to convert the values on existing instruments to a stream of digital values for control, and monitoring without the obligation to change, upgrade or tamper with them. During use, the image on an instrument face can become distorted for various reasons including dirt, noise, ambient lighting, camera issues and image processing artifacts. Recovering the underlying readings in real time is accomplished here using the same technique used for the recognition of handwriting - a deep learning convolutional neural network (CNN). This work defines a versatile result with wide applicability and an unexpected sampling benefit. Rather than remodel the world for robots with its concomitant time and expense, we enable robots to adapt to the real world.

ABSTRACT. The latest study by scientists from the International Agency for Research on Cancer (IARC) and partners forecasts that the number of new cases of cancer per year will increase by more than 50% from 2020 to 2040. Every fifth Americans will be diagnosed with cancer in their lifetime. Overexposure to UV light is the most common cause of skin cancer. It is very difficult to detect it in its early stages due to its apparent similarities. The ability to classify and detect skin cancer precisely has been made possible by recent technological advancements and enhanced research methodologies; however, there is still a lot to be done. In this paper, a deep learning technique for Skin cancer detection and classification is presented. This model has demonstrated State of the Art (SOTA) results in terms of accuracy and Frame per second and requires the least features as compared to previously used techniques such as ResNet, and AlexNet. The code is available at: https://github.com/RizwanMunawar/skincancer-binary-classification-computer-vision

ABSTRACT. This paper aims to present the software engineering process for the TrekkAid application that was developed to improve access to and safety within health and wellness green spaces. The main goal is to bring balance to the disproportion experienced by marginalized communities that have had barriers whether educational, discriminatory, socioeconomic, or otherwise affecting their abilities to utilize green spaces that are freely available. TrekkAid was designed and developed in response to the National Health Foundation’s call to action which seeks to improve accessibility of green spaces for individuals in these communities that have had disproportionately low representation. The TrekkAid mobile application was developed with modern software engineering architecture and principles. The application utilizes Cloud capabilities and services that work seamlessly on various platforms. The application also takes advantage of features and functionality which are readily available in mobile devices. The architecture and the various technologies are discussed in section three of the paper. In creating this application, the aim was a design and set of capabilities that would encourage individuals by equipping them with an interactive mobile experience which serves to educate them as well as allow them to navigate green spaces safely and more comfortably. Though this application can be used by anyone, special attention is placed on having an experience that is user-friendly for those in marginalized communities that have been less involved in green space activities.

ABSTRACT. Human Trafficking is one of the most lucrative activities and illegal businesses in the world. This activity got worst after the COVID situation, where people stay more at home connected to the internet all the time. Online predators are finding more opportunities to take advantage of the situation. The use of the internet is growing, but together the illegal activities are growing as well. Developing tools and investigation of these activities never have been so necessary for the last few years. The function of cyber activities behind this is to mitigate the actions of online predators and preserve the life and integrity of the population.

ABSTRACT. Gamified learning provides a fun, competitive environment for students to be rewarded to learn. Computer programming courses can be difficult for students as programming and problem solving require lots of practice and exposure. A gamified approach to such material would allow students to learn for fun and focus less on the frustrations associated with problem solving. However, teachers who want to incorporate games into their programming courses may feel lost on how to get started with the plethora of frameworks and games that they must learn to use.

In this paper, we provide information on a minimalist gamified approach for teachers to use in their programming courses. Students participate in a one hour competition on a biweekly basis to exercise their problem solving ability on easy, medium, and hard problems for various concepts. A leaderboard is used to track student scores. Students who complete a single problem are given a participation grade of 100 while lead competitors receive additional perks. We intend to analyze the effectiveness of a minimalist approach to gamification in learning programming and problem solving versus more traditional classroom methods.