Computer Engineering Commons^™

K-Perm: Personalized Response Generation Using Dynamic Knowledge Retrieval And Persona-Adaptive Queries, Kanak Raj, Kaushik Roy, Vamshi Bonagiri, Priyanshul Govil, Krishnaprasad Thirunarayan, Raxit Goswami, Manas Gaur

Publications

Personalizing conversational agents can enhance the quality of conversations and increase user engagement. However, they often lack external knowledge to tend to a user’s persona appropriately. This is particularly crucial for practical applications like mental health support, nutrition planning, culturally sensitive conversations, or reducing toxic behavior in conversational agents. To enhance the relevance and comprehensiveness of personalized responses, we propose using a two-step approach that involves (1) selectively integrating user personas and (2) contextualizing the response with supplementing information from a background knowledge source. We develop K-PERM (Knowledge-guided PErsonalization with Reward Modulation), a dynamic conversational agent that combines these elements. …

Tutorial: Knowledge-Infused Artificial Intelligence For Mental Healthcare, Kaushik Roy

Publications

Artificial Intelligence (AI) systems for mental healthcare (MHCare) have been ever-growing after realizing the importance of early interventions for patients with chronic mental health (MH) conditions. Social media (SocMedia) emerged as the go-to platform for supporting patients seeking MHCare. The creation of peer-support groups without social stigma has resulted in patients transitioning from clinical settings to SocMedia supported interactions for quick help. Researchers started exploring SocMedia content in search of cues that showcase correlation or causation between different MH conditions to design better interventional strategies. User-level Classification-based AI systems were designed to leverage diverse SocMedia data from various MH conditions, …

Personalized Bayesian Inference For Explainable Healthcare Management And Intervention, Utkarshani Jaimini, Krishnaprasad Thirunaravan, Maninder Kalra, Robin Dawson, Amit Sheth

Publications

Chronic healthcare conditions such as Asthma re- quires constant monitoring and managing of symptoms and their triggers for better quality of life. Each asthma patient reacts very differently to potential triggers. Hence, there is a need to develop a explainable personalized framework for each patient to capture susceptibility to asthma triggers. We developed a personalized knowledge-based probabilistic model to predict asthma exacerbation for different environmental factors utilizing patient generated health data from pediatric asthma patients. Further, the personalized model provides a metric, called Health Coefficient, to quantify the health of a patient for varying environmental factors. We demonstrate the predictive …

Causal Neuro-Symbolic Ai: A Synergy Between Causality And Neuro-Symbolic Methods, Utkarshani Jaimini, Cory Henson, Amit Sheth

Publications

Causal Neuro-Symbolic AI combines the benefits of causality with Neuro-Symbolic Artificial Intelligence (NeSyAI). More specifically, it (1) enriches NeSyAI systems with explicit representations of causality, (2) integrates causal knowledge with domain knowledge, and (3) enables the use of NeSyAI techniques for causal AI tasks. The explicit causal representation yields insights that predictive models may fail to analyze from observational data. It can also assist people in decision-making scenarios where discerning the cause of an outcome is necessary to choose among various interventions.

Ontolog Summit 2024 Talk Report: Healthcare Assistance Challenges-Driven Neurosymbolic Ai, Kaushik Roy

Publications

Although Artificial Intelligence technology has proven effective in providing healthcare assistance by analyzing health data, it still falls short in supporting decision-making. This deficiency largely stems from the predominance of opaque neural networks, particularly in mental health care AI applications, which raise concerns about their unpredictable and unverifiable nature. This skepticism hinders the transition from information support to decision support. This presentation will explore neurosymbolic approaches that combine neural networks with symbolic control and verification mechanisms. These approaches aim to unlock AI’s full potential by enhancing information analysis and decision-making support for healthcare assistance1.

A Comprehensive Survey On Rare Event Prediction, Chathurangi Shyalika Jayakody Kankanamalage, Ruwan Wickramarachchi, Amit Sheth

Publications

Rare event prediction involves identifying and forecasting events with a low probability using machine learning (ML) and data analysis. Due to the imbalanced data distributions, where the frequency of common events vastly outweighs that of rare events, it requires using specialized methods within each step of the ML pipeline, i.e., from data processing to algorithms to evaluation protocols. Predicting the occurrences of rare events is important for real-world applications, such as Industry 4.0, and is an active research area in statistics and ML. This paper comprehensively reviews the current approaches for rare event prediction along four dimensions: rare event data, …

Dynamic Modeling And Control Of A Solid State Semiconductor-Based Transformer, Microgrid And Storage Systems, Rubén Darío Viñán-Velasco

Department of Electrical and Computer Engineering: Dissertations, Theses, and Student Research

Smart Grids are power grid models designed with the idea of including the growing new technologies, from generation to storage devices, and are a response to the growing demands from consumers and the presence of electronic components being commonplace in the modern devices. The design requires a dynamic alternative in order to build an independent grid that can also work in cooperation with other micro-grids and the power grid in an integrated way. Smart-grids present several advantages over the traditional power grid scheme, but the economic costs of the components required to implement smart-grids is currently a great limitation. This …

Evaluating The Performance Of 5g Nr In Indoor Environments: An Experimental Study, Bikash Chandra Singh, Rafael Diaz, Sachin Shetty

School of Cybersecurity Faculty Publications

The 5G wireless standard has emerged as a trans-formative technology with the potential to revolutionize various industries by providing enhanced connectivity and communication capabilities. This advanced standard offers a diverse range of applications, including Ultra-Reliable Low-Latency Communication (URLLC), Enhanced Mobile Broadband (eMBB), and Massive Machine Type Communication (mMTC). In this Scientific research paper, we present a comprehensive analysis of the performance and capabilities of a deployed indoor 5G network in a controlled laboratory environment. The experimental setup comprises an Amarisoft Callbox, serving as the 5G core, along with a Remote Radio Head (RRH) and user equipment (UEs). Our primary objective …

Reinventing Integrated Photonic Devices And Circuits For High Performance Communication And Computing Applications, Venkata Sai Praneeth Karempudi

Theses and Dissertations--Electrical and Computer Engineering

The long-standing technological pillars for computing systems evolution, namely Moore's law and Von Neumann architecture, are breaking down under the pressure of meeting the capacity and energy efficiency demands of computing and communication architectures that are designed to process modern data-centric applications related to Artificial Intelligence (AI), Big Data, and Internet-of-Things (IoT). In response, both industry and academia have turned to 'more-than-Moore' technologies for realizing hardware architectures for communication and computing. Fortunately, Silicon Photonics (SiPh) has emerged as one highly promising ‘more-than-Moore’ technology. Recent progress has enabled SiPh-based interconnects to outperform traditional electrical interconnects, offering advantages like high bandwidth density, …

Fake News Detection In Online Platforms, Elena Shushkevich

Doctoral

This thesis presents research conducted during a Ph.D. program at Technological University Dublin from 2020 to 2024. The objective of this research is to develop and evaluate effective methods for detecting and classifying fake news in social media and press, addressing the critical issue of misinformation in the digital age. The relevance of this study is underscored by the increasing prevalence of fake news and its potential societal impact, emphasizing the importance of advanced tools for identifying and mitigating misinformation.

Detection Of Tooth Position By Yolov4 And Various Dental Problems Based On Cnn With Bitewing Radiograph, Kuo Chen Li, Yi-Cheng Mao, Mu-Feng Lin, Yi-Qian Li, Chiung-An Chen, Tsung-Yi Chen, Patricia Angela R. Abu

Department of Information Systems & Computer Science Faculty Publications

Periodontitis is a high prevalence dental disease caused by bacterial infection of the bone that surrounds the tooth. Early detection and precision treatment can prevent more severe symptoms such as tooth loss. Traditionally, periodontal disease is identified and labeled manually by dental professionals. The task requires expertise and extensive experience, and it is highly repetitive and time-consuming. The aim of this study is to explore the application of AI in the field of dental medicine. With the inherent learning capabilities, AI exhibits remarkable proficiency in processing extensive datasets and effectively managing repetitive tasks. This is particularly advantageous in professions demanding …

Enhancing Robotic Exploration Through Semantically-Guided Sampling Strategies, Christopher Alexander Arend Tatsch

Graduate Theses, Dissertations, and Problem Reports (ETD)

From space and deep-sea exploration to disaster response and environmental monitoring, autonomous robots are essential for advancing science, improving safety, and addressing critical challenges. This dissertation introduces a novel open-source strategy for autonomous robotic exploration: the Semantically-Guided Exploration (SGE) framework. Designed for ground vehicles, SGE integrates semantic understanding into the autonomous exploration process, improving decision-making in complex environments. Specifically, the proposed sampling-based approach uses the information from the semantic segmentation of RGB images and depth images to guide the robot's selection of exploration goals. This method enables the robot to steer away from potential dangers such as large rocks and …

Observability-Aware Path Planning For Autonomous Navigation, Raymond Brink Neistat

Open Access Master's Theses

Terrain-Aided Navigation (TAN) is a popular method of localization for GPS-denied vehicles, particularly in the marine domain. There are many ways to perform TAN in a marine setting, such as Bathymetric SLAM (BSLAM) and Bayesian Filtering with the aid of an a priori map. These techniques have been studied extensively, but show an overall lack of rigorous observability analyses. Without nonlinear observability analyses, TAN practitioners do not have an analytical indicator to know which areas of terrain will provide opportunities for the best localization performance. This thesis reviews current developments in the endeavors of nonlinear observability analyses as well as …

Smart Dog Door, Andrew Shetler, William Boissoneault, Jacob Stump, Benjamin Charlson

Williams Honors College, Honors Research Projects

With modern life getting busier and more complex, many have turned to autonomous assistance for taking care of pets. The objective of this project is to design and prototype a device and application that will allow only certain dogs to enter and exit through a dog door while making the owners aware of their dog's activity. The dog door will utilize wireless communication from the dog to the door to control the status of the door’s lock and update the database that will send updates to the application. The application will keep track of when the dog enters and exits …

Midi Recorder And Learning Tool, Caden Dees, Saikishore Gowrishankar, Cameron Johnson, Benjamin Bolyard

Williams Honors College, Honors Research Projects

The project's goal is to create a device that can record MIDI files from any standard 88-key piano as well as teach piano learners how to play songs.

Autonomous Damage And Structure Scanning Drone, Natasha Ninan, Amber Long, Lee Nestor, Emmanuel Jensen

Williams Honors College, Honors Research Projects

Remote damage analysis plays a crucial role in lowering the risk associated with human presence at dangerous sites. This project focuses on developing a system for remote structural examination using photogrammetric techniques and analysis. The system utilizes a drone equipped with cameras for photogrammetry, and LiDARs for navigation. This setup can enable efficient structural model generation with the collection of images from multiple points. A novel structural analysis is performed to detect potential damage or points of failure in the structure.

Key components include a teleoperated drone, a software pipeline for photogrammetric analysis, collision protection mechanisms, and a user interface. …

Autonomous Basketball Court Creation Robot, Bryce Haldeman, Tyler Gray, Dalon Vura

Williams Honors College, Honors Research Projects

The Autonomous Basketball Court Outlining System presents a comprehensive solution for precision court marking. Powered by a 24V lithium-ion battery and driven by a single ST microcontroller, the system autonomously marks the outline of a half basketball court using predefined algorithms. User-friendly features include easy loading of marking material, actuated by gravity or a small servo motor depending on material of choice, ensuring intuitive operation. Safety is prioritized, with the servo motor eliminating high-pressure concerns, and the system maintains a controlled speed accounting for user well-being. Two step and direction servo motors enable accurate linear displacement, facilitating straight lines, and …

Zero-Shot Cross-Lingual Pos Tagging For Filipino, Jimson Paulo Layacan, Isaiah Edri W. Flores, Katrina Bernice M. Tan, Ma. Regina Justina Estuar, Jann Railey E. Montalan, Marlene M. De Leon

Department of Information Systems & Computer Science Faculty Publications

Supervised learning approaches in NLP, exemplified by POS tagging, rely heavily on the presence of large amounts of annotated data. However, acquiring such data often requires significant amount of resources and incurs high costs. In this work, we explore zero-shot cross-lingual transfer learning to address data scarcity issues in Filipino POS tagging, particularly focusing on optimizing source language selection. Our zero-shot approach demonstrates superior performance compared to previous studies, with top-performing fine-tuned PLMs achieving F1 scores as high as 79.10%. The analysis reveals moderate correlations between cross-lingual transfer performance and specific linguistic distances–featural, inventory, and syntactic–suggesting that source languages with …

Efficient Connectivity Management And Path Planning For Iot And Uav Networks, Amirahmad Chapnevis

Theses and Dissertations

This dissertation explores how to better manage resources in mobile networks, especially for enhancing the performance of Unmanned Aerial Vehicles (UAV)-supported IoT networks. We explored ways to set up a flexible communication architecture that can handle large IoT deployments by making good use of mobile core network resources like bearers and data paths. We developed strategies that meet the needs of IoT networks and enhance network performance. We also developed and tested a system that combines traffic from several mobile devices that use the same user identity and network resources within the core mobile network. We used everyday smartphones, SIM …

Implementing Associative Learning Using Neuromorphic Robot, Vinay Kumar Pillalamarri

Dissertations, Master's Theses and Master's Reports

Associative learning, a key cognitive process seen across the animal kingdom, enables organisms to form connections between stimuli and adapt their behaviors based on past experiences. A particularly powerful example is fear conditioning, where animals learn to associate a neutral stimulus with an aversive one, allowing them to predict and avoid potential threats. Inspired by this mechanism, this project implements associative learning on an unmanned ground vehicle (UGV) to develop adaptive behavior through neuromorphic principles. Utilizing Nengo for neural modeling, the UGV learns to associate visual (red color) and tactile (vibration) stimuli through Hebbian learning, a biologically inspired synaptic adaptation …

The Integration Of Neuromorphic Computing In Autonomous Robotic Systems, Md Abu Bakr Siddique

Dissertations, Master's Theses and Master's Reports

Deep Neural Networks (DNNs) have come a long way in many cognitive tasks by training on large, labeled datasets. However, this method has problems in places with limited data and energy, like when planetary robots are used or when edge computing is used [1]. In contrast to this data-heavy approach, animals demonstrate an innate ability to learn by communicating with their environment and forming associative memories among events and entities, a process known as associative learning [2-4]. For instance, rats in a T-maze learn to associate different stimuli with outcomes through exploration without needing labeled data [5]. This learning paradigm …

A Novel Processor Architecture Implementing The Stacked Error Diffusion Algorithm And Its Zynq-Based Realization, Qishi Hu

Theses and Dissertations--Electrical and Computer Engineering

Digital halftoning reproduces continuous-tone images using patterns of black and white dots, while multitoning extends this concept by incorporating inks with intermediate intensities. These techniques are extensively utilized in the printing industry to accommodate the limited range of inks available in printers. Stacked error diffusion is a high-quality multitoning algorithm that adheres to the blue-noise dithering standard. This thesis research studies the potential parallelism inherent in the algorithm and introduces the design of a novel processor architecture optimized for efficient execution. The architecture is realized on an FPGA development board featuring a Zynq SoC. Additionally, the hardware prototype can also …

Information Access For Infrastructurally-Challenged Environments And Beyond Through Mutually Aware Spectrum Sharing Technologies, Karyn Doke

Electronic Theses & Dissertations (2024 - present)

The Radio Frequency (RF) spectrum is scarce and to make it available for new mobile wireless services, regulators are forced to re-allocate spectrum from existing services or develop mechanisms to share spectrum with new entries. Television White Space (TVWS) and Citizen Broadband Radio Service (CBRS) are two examples of recently commercialized spectrum sharing technologies. TVWS enables sharing among fixed wireless broadband technologies (secondary users) and terrestrial TV broadcast services (primary users). CBRS enables spectrum sharing among 5G/LTE (secondary users) and naval radar (primary users). With both technologies, a central database determines when it is safe for secondary users to operate …

Exploratory Prompting Of Large Language Models To Act As Co-Pilots For Augmenting Business Process Work In Document Classification, Jose Ramon Ilagan, Joseph Benjamin R. Ilagan, Claire Louisse Basallo, Zachary Matthew Alabastro

Quantitative Methods and Information Technology Faculty Publications

Businesses deal with different types of documents containing unstructured documents. The data in these documents must be converted into digital forms other automated systems could only process. One generic use case is document classification, which usually involves manual transformation due to human understanding needed in the process. These documents go beyond those generated through regular business transactions and operations and also include web-based content such as online news, blogs, e-mails, and various digital libraries. Recent developments in robotic process automation (RPA) and artificial intelligence (AI) aim to automate the otherwise expensive, time-consuming, and repetitive manual steps. Through more powerful natural …

Evaluating The Impact Of Perceptual Loss In Generative Adversarial Models And Diffusion Models For Document Image Enhancement, Farzaneh Karimpour

Electronic Theses and Dissertations

Documents often suffer from various types of degradation which make them difficult to read and restrict OCR performance. This study investigates the effectiveness of perceptual loss in enhancing document image cleanup by comparing a GAN-based model and a diffusion model. In our experiments, we utilized the DE-GAN model as a GAN-based model and the NAF-DPM model as a diffusion model, both enhanced by incorporating perceptual loss. We then compared the results of both models and evaluated them by using the DIBCO 2013, DIBCO 2017, and H-DIBCO 2018 datasets revealed that our approach consistently outperforms existing state-of-the-art methods. Results showed that …

Comparative Analysis Of Deep Learning-Based Anomaly Detection Models For Gps Spoofing Detection, Hasan Mirzakhaninafchi

Electronic Theses and Dissertations

As autonomous vehicles (AVs) become integral to modern transportation, their susceptibility to cyber-attacks, particularly GPS spoofing, presents a serious security threat. This study addresses these challenges by applying a suite of deep learning models to enhance the detection of anomalous GPS signals. Focusing on autoencoder-based architectures, the proposed models such as long short-term memory-based variational autoencoder (LSTM-VAE), LSTM-based autoencoder (LSTM-AE), multilayer perceptron-based variational autoencoder (MLP-VAE), MLP-based Autoencoder (MLPAE), Stacked LSTM-based variational autoencoder (Stacked-LSTM-VAE), stacked LSTM-based autoencoder (Stacked-LSTM-AE), memory-augmented-LSTM-VAE (Mem-LSTM-VAE), and time-series-anomaly-detection-generative-adversarial-networks (TadGAN) were trained exclusively on authentic GPS data. This unsupervised learning approach which used for the above-mentioned models enables …

An Integrated Hybrid P2p-Dr Networks For A Transactive Energy Market Platform Considering Electricity Network Constraints, Sheroze Liaquat

Electronic Theses and Dissertations

No abstract provided.

An Fpga-Based Eit System For Deep Space Medical Imaging, Kendall R. Farnham

Dartmouth College Ph.D Dissertations

Dangers associated with high radiation and microgravity exposure in space are critical challenges inhibiting us from exploring deep space and pursuing long-duration missions, as current medical systems are unable to monitor, diagnose, or treat tissue injury within physical spacecraft constraints and communication limits. Ultrasound (US) is the current imaging system used on the International Space Station, but this technology relies on telemedical support (or onboard artificial intelligence/autonomous capabilities) for both operation and diagnosis, posing challenges for crews isolated in deep space. Electrical impedance tomography (EIT) is a non-invasive, non-ionizing technology that produces images of the electrical properties of tissues and …

Neurosymbolic Value-Inspired Ai (Why, What, And How), Amit Sheth, Kaushik Roy

Publications

The rapid progression of Artificial Intelligence (AI) systems, facilitated by the advent of Large Language Models (LLMs), has resulted in their widespread application to provide human assistance across diverse industries. This trend has sparked significant discourse centered around the ever-increasing need for LLM-based AI systems to function among humans as part of human society, sharing human values, especially as these systems are deployed in high-stakes settings (e.g., healthcare, autonomous driving, etc.). Towards this end, neurosymbolic AI systems are attractive due to their potential to enable easy-tounderstand and interpretable interfaces for facilitating valuebased decision-making, by leveraging explicit representations of shared values. …

Exploring Alternative Approaches To Language Modeling For Learning From Data And Knowledge, Yuxin Zi, Kaushik Roy, Vignesh Narayanan, Amit Sheth

Publications

Despite their wide applications to language understanding tasks, large language models (LLMs) still face challenges such as hallucinations - the occasional fabrication of information, and alignment issues - the lack of associations with human-curated world models (e.g., intuitive physics or common-sense knowledge). Additionally, the black-box nature of LLMs makes it highly challenging to train them meaningfully in order to achieve a desired behavior. Specifically, the attempt to adjust LLMs’ concept embedding spaces can be highly intractable, which involves analyzing the implicit impact on LLMs’ numerous parameters and the resulting inductive biases. This paper proposes a novel architecture that wraps powerful …