Digital Commons Network^™

Explaining Sequences Of Actions In Multi-Agent Deep Reinforcement Learning Models, Phyo Wai Khaing, Minghong Geng, Shubham Pateria, Budhitama Subagdja, Ah-Hwee Tan

Research Collection School Of Computing and Information Systems

This paper introduces a method to explain MADRL agents’ behaviors by abstracting their actions into high-level strategies. Particularly, a spatio-temporal neural network model is applied to encode the agents’ sequences of actions as memory episodes wherein an aggregating memory retrieval can generalize them into a concise abstract representation of collective strategies. To assess the effectiveness of our method, we applied it to explain the actions of QMIX MADRL agents playing a StarCraft Multi-agent Challenge (SMAC) video game. A user study on the perceived explainability of the extracted strategies indicates that our method can provide comprehensible explanations at various levels of …

Policy-Based Remote User Authentication From Multi-Biometrics, Yangguang Tian, Yingjiu Li, Robert H. Deng, Guomin Yang, Nan Li

Research Collection School Of Computing and Information Systems

In this paper, we introduce the first generic framework of policy-based remote user authentication from multiple biometrics. The proposed framework allows an authorized user to remotely authenticate herself to an authentication server using her multiple biometrics, which enhances both the security and usability of user authentications. The authentication server approves a user's authentication request if and only if the user's multiple biometrics satisfies an authentication policy. In particular, the authentication policy can be dynamically updated to satisfy different security and usability requirements in practice. We implement an instantiation of the proposed framework and report its performance under various authentication policies.

Rule-Guided Counterfactual Explainable Recommendation, Yinwei Wei, Xiaoyang Qu, Xiang Wang, Yunshan Ma, Liqiang Nie, Tat‑Seng Chua

Research Collection School Of Computing and Information Systems

To empower the trust of current recommender systems, the counterfactual explanation (CE) method is adopted to generate the counterfactual instance for each input and take their changes causing the different outcomes as the explanation. Although promising results have been achieved by existing CE-based methods, we propose to generate the attribute-oriented counterfactual explanation. Different from them, we aim to generate the counterfactual instance by performing the intervention on the attributes, and then build an attribute-oriented counterfactual explainable recommender system. Considering the correlation and categorical values of attributes, how to efficiently generate the reliable counterfactual instances on the attributes challenges us. To …

Learning To Generate Explainable Stock Predictions Using Self‑Reflective Large Language Models, Kelvin J.L. Koa, Yunshan Ma, Ritchie Ng, Tat‑Seng Chua

Research Collection School Of Computing and Information Systems

Explaining stock predictions is generally a difficult task for traditional non-generative deep learning models, where explanations are limited to visualizing the attention weights on important texts. Today, Large Language Models (LLMs) present a solution to this problem, given their known capabilities to generate human-readable explanations for their decision-making process. However, the task of stock prediction remains challenging for LLMs, as it requires the ability to weigh the varying impacts of chaotic social texts on stock prices. The problem gets progressively harder with the introduction of the explanation component, which requires LLMs to explain verbally why certain factors are more important …

Attribute-Hiding Fuzzy Encryption For Privacy-Preserving Data Evaluation, Zhenhua Chen, Luqi Huang, Guomin Yang, Willy Susilo, Xingbing Fu, Xingxing Jia

Research Collection School Of Computing and Information Systems

Privacy-preserving data evaluation is one of the prominent research topics in the big data era. In many data evaluation applications that involve sensitive information, such as the medical records of patients in a medical system, protecting data privacy during the data evaluation process has become an essential requirement. Aiming at solving this problem, numerous fuzzy encryption systems for different similarity metrics have been proposed in literature. Unfortunately, the existing fuzzy encryption systems either fail to achieve attribute-hiding or achieve it, but are impractical. In this paper, we propose a new fuzzy encryption scheme for privacy-preserving data evaluation based on overlap …

Vibmilk: Non-Intrusive Milk Spoilage Detection Via Smartphone Vibration, Yuezhong Wu, Wei Song, Yanxiang Wang, Dong Ma, Weitao Xu, Mahbub Hassan, Wen Hu

Research Collection School Of Computing and Information Systems

Quantifying the chemical process of milk spoilage is challenging due to the need for bulky, expensive equipment that is not user-friendly for milk producers or customers. This lack of a convenient and accurate milk spoilage detection system can cause two significant issues. First, people who consume spoiled milk may experience serious health problems. Secondly, milk manufacturers typically provide a “best before” date to indicate freshness, but this date only shows the highest quality of the milk, not the last day it can be safely consumed, leading to significant milk waste. A practical and efficient solution to this problem is proposed …

Compositional Policy Learning In Stochastic Control Systems With Formal Guarantees, Dorde Zikelic, Mathias Lechner, Abhinav Verma, Krishnendu Chatterjee, Thomas A. Henzinger

Research Collection School Of Computing and Information Systems

Reinforcement learning has shown promising results in learning neural network policies for complicated control tasks. However, the lack of formal guarantees about the behavior of such policies remains an impediment to their deployment. We propose a novel method for learning a composition of neural network policies in stochastic environments, along with a formal certificate which guarantees that a specification over the policy's behavior is satisfied with the desired probability. Unlike prior work on verifiable RL, our approach leverages the compositional nature of logical specifications provided in SPECTRL, to learn over graphs of probabilistic reach-avoid specifications. The formal guarantees are provided …

Learning, Optimizing, And Simulating Fermions With Quantum Computers, Andrew Zhao

Physics & Astronomy ETDs

Fermions are fundamental particles which obey seemingly bizarre quantum-mechanical principles, yet constitute all the ordinary matter that we inhabit. As such, their study is heavily motivated from both fundamental and practical incentives. In this dissertation, we will explore how the tools of quantum information and computation can assist us on both of these fronts. We primarily do so through the task of partial state learning: tomographic protocols for acquiring a reduced, but sufficient, classical description of a quantum system. Developing fast methods for partial tomography addresses a critical bottleneck in quantum simulation algorithms, which is a particularly pressing issue for …

Flipped Classroom For Linear Algebra At Undergraduate Level, M. Thulasidas

Research Collection School Of Computing and Information Systems

In this article, we describe our experience in developing an undergraduate Linear Algebra course tailored to highlight its relevance and applicability in Computer Science. Over the course of three years, the course transitioned from a traditional direct-instruction format to a flipped-classroom design, resulting in positive student learning outcomes. This article covers the course design philosophy, its syllabus, learning objectives, and the incorporation of both quantitative and qualitative student feedback in shaping the course. Furthermore, the article shares the insights gleaned from our experience, which can serve as best practices for instructors aiming to deliver a successful Linear Algebra course for …

The Grader: A Grading Assistant For Lab Tests And A Teaching Tool, M. Thulasidas, David Lo

Research Collection School Of Computing and Information Systems

This article presents the design and implementation of the Grader, a grading assistant application deployed for a Web Application Development course at our school. The Grader is equipped to handle various logistical aspects of lab tests, including file management, consistent application of rubrics, and auto-grading of questions with test cases. Additionally, it incorporates heuristic rules to detect cheating attempts. We anticipate that the Grader will find widespread utility in programming courses where lab tests serve as summative assessments. Developed within the same programming environment taught in the class, the Grader also serves as a pedagogical tool, demonstrating to students a …

Robust Auto-Scaling With Probabilistic Workload Forecasting For Cloud Databases, Haitian Hang, Xiu Tang, Jianling Sun, Lingfeng Bao, David Lo, Haoye Wang

Research Collection School Of Computing and Information Systems

Auto-scaling is crucial for achieving elasticity in cloud databases as well as other cloud systems. Predictive auto-scaling, which leverages forecasting techniques to adjust resources based on predicted workload, has been widely adopted. However, the inherent inaccuracy of forecasting presents a significant challenge, potentially causing resource under-provisioning. To address this challenge, we propose robust predictive auto-scaling that considers the uncertainty in forecasts. Unlike previous predictive approaches that rely on single-valued forecasts, we leverage probabilistic forecasting techniques to generate quan-tile forecasts, providing a more comprehensive understanding of the potential future workloads. By formulating the auto-scaling problem as a robust optimization problem, we …

Automatic Grading Of Short Answers Using Large Language Models In Software Engineering Courses, Nguyen Binh Duong Ta, Yi Meng Chai

Research Collection School Of Computing and Information Systems

Short-answer based questions have been used widely due to their effectiveness in assessing whether the desired learning outcomes have been attained by students. However, due to their open-ended nature, many different answers could be considered entirely or partially correct for the same question. In the context of computer science and software engineering courses where the enrolment has been increasing recently, manual grading of short-answer questions is a time-consuming and tedious process for instructors. In software engineering courses, assessments concern not just coding but many other aspects of software development such as system analysis, architecture design, software processes and operation methodologies …

Representation Learning For Generative Models With Applications To Healthcare, Astronautics, And Aviation, Van Minh Nguyen

Theses and Dissertations

This dissertation explores applications of representation learning and generative models to challenges in healthcare, astronautics, and aviation.

The first part investigates the use of Generative Adversarial Networks (GANs) to synthesize realistic electronic health record (EHR) data. An initial attempt at training a GAN on the MIMIC-IV dataset encountered stability and convergence issues, motivating a deeper study of 1-Lipschitz regularization techniques for Auxiliary Classifier GANs (AC-GANs). An extensive ablation study on the CIFAR-10 dataset found that Spectral Normalization is key for AC-GAN stability and performance, while Weight Clipping fails to converge without Spectral Normalization. Analysis of the training dynamics provided further …

Automated Composition Of Multivariable Scientific Workflows Considering Scientific Assumptions, Raul Alejandro Vargas Acosta

Open Access Theses & Dissertations

Many ground-breaking scientific experiments require the execution of multiple complex scientific computations. Thus, scientific workflows (i.e., a sequence of scientific computations) have received significant attention, more specifically, the automated composition of scientific workflows. Scientific workflows that repurpose data may have unique scientific assumptions that need to be considered when composing a workflow. Workflow composition tools have enabled a wider range of stakeholders (e.g., policymakers, the general public, and researchers) to create and execute workflows; however, domain expertise is still required for these tasks. The overarching goal of this work is to further improve the automatic composition of scientific workflows by …

Satellite Image Analysis And Sidewalk Classification Using Deep Learning Models, Subeksha Khanal

Master's Theses

Lack of sidewalk pavement can contribute to pedestrian fatalities and injuries in the USA. Although many researchers have conducted research on sidewalk detection, there are not many publicly available datasets that we would work on for sidewalk classification. In this study, I conducted classification tasks using pretrained CNN models including VGG16 and ResNet50. I extended these models by adding custom layers at the top of pretrained layers, employing various techniques to improve the classification accuracy.

The dataset comprises 4,731 images of sidewalk based on occlusion levels, including images where the sidewalk is visible from overhead and instances where sidewalk is …

A Review Of Student Attitudes Towards Keystroke Logging And Plagiarism Detection In Introductory Computer Science Courses, Caleb Syndergaard

All Graduate Theses and Dissertations, Fall 2023 to Present

The following paper addresses student attitudes towards keystroke logging and plagiarism prevention measures. Specifically, the paper concerns itself with changes made to the “ShowYourWork” plugin, which was implemented to log the keystrokes of students in Utah State University’s introductory Computer Science course, CS1400. Recent work performed by the Edwards Lab provided insights into students’ feelings towards keystroke logging as a measure of deterring plagiarism. As a result of that research, we have concluded that measures need to be taken to enable students to have more control over their data and assist students to feel more comfortable with keystroke logging. This …

Inferring A Hierarchical Input Type For An Sql Query, Santosh Aryal

All Graduate Theses and Dissertations, Fall 2023 to Present

SQL queries are a common method to retrieve information from databases, much like asking a detailed question and getting a precise answer. Plug-and-play queries simplify the process of querying. In a Plug-and-play SQL query a programmer sketches the shape of the input to the query as a hierarchy. But the programmer could make a mistake in specifying the hierarchy and it takes programmer time and effort to specify the hierarchy. A better solution is to automatically infer the hierarchy from a query. This thesis presents a system to infer a hierarchical input type for an SQL query. We consider two …

Empowering Graphics: A Distributed Rendering Architecture For Inclusive Access To Modern Gpu Capabilities, Taylor Anderson

All Graduate Theses and Dissertations, Fall 2023 to Present

Modern rendering software requires powerful GPUs with the latest hardware features in order to utilize all of the newest rendering techniques. Many users do not have access to this hardware, and rely on remote server farms or reduced performance to achieve usable results. In this thesis, the software is designed and created to allow for a user to share the resources of their computer with another, modeling a split-screen setup like was common in the past, but without requiring users to be in the same location.

By designing the software from the ground up to support this, instead of adding …

A Comparative Analysis Of Field Electron Emission From Carbon Black Embedded Within Insulated Copper Hollowed Wires And Glass Tubes, Hatem A. Al-Braikat, Ahmad M D Jaber, Adel M. Abuamr, Mazen A. Madanat, Aseel A. Al-Jbarart, M-Ali H. Al-Akhras, Marwan S. Mousa

Karbala International Journal of Modern Science

In this study, two different methods are used to investigate carbon black as a cold field electron emitter. The first method is to incorporate carbon black into a specially designed insulated copper hollowed wire. The wire has a cup-shaped structure created by electrochemical etching. The second method involves the incorporation of carbon black into narrow glass tubes. A Comparative analyses is carried out to evaluate the effectiveness of each method. To evaluate the performance of the samples, the current-voltage characteristics will be examined using field electron microscopes. This analysis will provide an understanding of the emission of the carbon black …

Advancing Mobile Sensing In Dynamic Environments, Weichen Wang

Dartmouth College Ph.D Dissertations

This thesis presents a comprehensive exploration of enhancing mobile sensing capabilities to address various aspects of human behavior, mental health, personality, social functioning and beyond. We redesign the StudentLife app to improve its sensing efficiency and dependability, enabling support for multi-year-long studies. By adopting new app design, this study addresses the technical challenges of continuous sensing and enhances system robustness. The work is organized into several key studies that collectively aim to expand the scope of mobile sensing in diverse and complex environments.

The first study broadens the scope of mobile sensing to assess personality traits, exploring the potential of …

Develop An Interactive Python Dashboard For Analyzing Ezproxy Logs, Andy Huff, Matthew Roth, Weiling Liu

Faculty and Staff Scholarship

This paper describes the development of an interactive dashboard in Python with EZproxy log data. Hopefully, this dashboard will help improve the evidence-based decision-making process in electronic resources management and explore the impact of library use.

Osint Research Studios: A Flexible Crowdsourcing Framework To Scale Up Open Source Intelligence Investigations, A. Mukhopadhyay, Sukrit Venkatagiri, K. Luther

Computer Science Faculty Works

Open Source Intelligence (OSINT) investigations, which rely entirely on publicly available data such as social media, play an increasingly important role in solving crimes and holding governments accountable. The growing volume of data and complex nature of tasks, however, means there is a pressing need to scale and speed up OSINT investigations. Expert-led crowdsourcing approaches show promise, but tend to either focus on narrow tasks or domains, or require resource-intense, long-term relationships between expert investigators and crowds. We address this gap by providing a flexible framework that enables investigators across domains to enlist crowdsourced support for discovery and verification of …

Priming Through Persuasion: Towards Secure Password Behavior, Rizu Paudel, Mahdi Nasrullah Al-Ameen

Computer Science Student Research

Users tend to create weak passwords even for the important accounts. The prior research shed light on user’s insecure password behavior, and why the interventions, including requirement specification (e.g., password composition policies) and feedback systems (e.g., password meters) fail in practice. To this end, we propose and evaluate the concept: priming-through-persuasion in the realm of secure password creation. In particular, we created visual designs, aimed at priming users about the repercussions of weak passwords before their password creation. We base our designs on two forms of persuasion methods: pathos and logos. Pathos appeals to people’s emotion in order to persuade …

“It’S Hard For Him To Make Choices Sometimes And He Needs Guidance”: Re-Orienting Parental Control For Children, Prakriti Dumaru, Hanieh Atashpanjeh, Mahdi Nasrullah Al-Ameen

Computer Science Student Research

With access to devices and online platforms at an increasingly younger age, it is essential to protect children from potential online risks. In our studies, we particularly focused the parents who have a child under 13 years of age. To this end, deriving from developmental psychology, we base our low-fidelity prototype design (termed as ‘treatment’) on a set of design principles focusing on open communication, instilling self-regulation, and availing granularity along with designs of existing Google’s parental control as a baseline. First, we conducted semi-structured interviews with 21 parents to understand their perceptions, including suggestions for design improvement within our …

Gmr-29 Identification Of Ai-Generated Images, Chris Foster, Joshua Brock, Harini Kottala, Srilatha Korrapati

C-Day Computing Showcase

With the quick rise of Artificial Intelligence (AI), generative AI models have greatly increased the volume and velocity of data creation. Among that data, AI-generated images have become a highly discussed topic, especially when discussing the potential dangers of these AI models. Due to these dangers, being able to distinguish AI-generated art from human-made art is becoming a necessity. Additionally, as these AI-models improve, it is becoming increasingly difficult for humans to determine whether art is AI-generated or human-made. This paper proposes the further exploration of the effectiveness of a current state of the art AI-image identification model.

Gmr-72 Ai-Based Discourse Analysis System (Adas) For Improved Stem Education, Varun Gottam

C-Day Computing Showcase

In the rapidly evolving fields of Artificial Intelligence and Natural Language Processing, significant opportunities have emerged to transform educational practices. Discourse analysis, particularly in science education, plays a critical role in fostering scientific thinking among students. However, the manual application of tools like the Classroom Discourse Analysis Tool is resource-intensive and impractical on a large scale. This abstract proposes the development of an AI-based Discourse Analysis System tailored for educational settings, designed to automate and enrich the analysis of classroom discourse. Leveraging the latest in Artificial Intelligence and Natural Language Processing, this web-based application will provide teachers nationwide with the …

Uc-85 Helpr: Helping Extrapolate Labels For Police Reports Using Large Language Models, Hailey N Walker, William A Stigall

C-Day Computing Showcase

Police officers spend many hours a week documenting their findings when reporting to a 911 call. There is so much detail in these reports that they remain an untapped resource for future data analytics by the police department. The reports are currently being analyzed by human experts and categorized into the following categories: “Substance Abuse”, “Mental Health”, “Domestic/Social”, “Nondomestic/Social”, and “Other”. To assist the experts and reduce the amount of time that is spent on reading and analyzing, we are proposing the use of large language models (LLMs) to tag police reports based on their content. Two models, Mistral-7B and …

Gpr-103 Personalized Pedagogy Through A Llm-Based Recommender System, Mourya Teja Kunuku, Bharath Y Yadla

C-Day Computing Showcase

The educational domain is undergoing transformation due to the incorporation of Artificial Intelligence (AI), Large Language Models (LLMs), and generative AI technologies, raising the need for educators to integrate cutting-edge technological advancements and methodologies into their teaching approaches. Pedagogical Design Patterns (PDPs) have become prominent for their role in sharing effective educational practices and narrowing the divide between academic research and actual teaching methods. Despite their potential, the lack of widely accessible resources and the scattered nature of publishing outlets pose significant barriers to the broad application of PDPS. To address this issue, we propose the application of large language …

Uc-48 Birding With Buddy, Lazare V Sawadogo, Ikhelowa E Adeji, Blake Graham, Zach Alpine, Troy C Sorrells

C-Day Computing Showcase

Birding with Buddy is an educational and entertaining immersive virtual 3D low-poly birdwatching to be experienced at the Carter Lake Nature Center to enable kids to embark on a quest to learn more about birds. Buddy the Beaver guides the user through different terrain types to identify diverse bird species with sounds. Integrate a bird identification system where players click on the binocular icon to switch to a binocular view. In this view, players can choose to Identify (multiple-choice) the correct bird, Hear the Call Again, or Consult a Field Guide. Featuring flippable pages with images and notable markings of …

Uc-69 Interactive Training Game Suite, Garrett J Perry, Joscelyn Cauley, Issabella Du, Sergiu Ursu, Rahaf Kokash

C-Day Computing Showcase

It is now common knowledge that simple lectures are not the most effective way for the average person to learn and retain knowledge. The Core of Engineers at the Warner Robins Air Logistic Center have tasked us to transform their PowerPoint presentations into interactive training games to improve comprehension, interaction and retention while saving time and logistical resources compared to giving a traditional lecture. We been tasked with creating game modules covering STINFO or what is or is not considered classified information, Records Management, and the No FEAR Act detailing whistleblower rights and protocols. Our team has developed a log …