Value-Based Subgoal Discovery And Path Planning For Reaching Long-Horizon Goals, Shubham Pateria, Budhitama Subagdja, Ah-Hwee Tan, Chai Quek

Research Collection School Of Computing and Information Systems

Learning to reach long-horizon goals in spatial traversal tasks is a significant challenge for autonomous agents. Recent subgoal graph-based planning methods address this challenge by decomposing a goal into a sequence of shorter-horizon subgoals. These methods, however, use arbitrary heuristics for sampling or discovering subgoals, which may not conform to the cumulative reward distribution. Moreover, they are prone to learning erroneous connections (edges) between subgoals, especially those lying across obstacles. To address these issues, this article proposes a novel subgoal graph-based planning method called learning subgoal graph using value-based subgoal discovery and automatic pruning (LSGVP). The proposed method uses a …

Anopas: Practical Anonymous Transit Pass From Group Signatures With Time-Bound Keys, Rui Shi, Yang Yang, Yingjiu Li, Huamin Feng, Hwee Hwa Pang, Robert H. Deng

Research Collection School Of Computing and Information Systems

An anonymous transit pass system allows passengers to access transport services within fixed time periods, with their privileges automatically deactivating upon time expiration. Although existing transit pass systems are deployable on powerful devices like PCs, their adaptation to more user-friendly devices, such as mobile phones with smart cards, is inefficient due to their reliance on heavy-weight operations like bilinear maps. In this paper, we introduce an innovative anonymous transit pass system, dubbed Anopas, optimized for deployment on mobile phones with smart cards, where the smart card is responsible for crucial lightweight operations and the mobile phone handles key-independent and time-consuming …

An Llm-Assisted Easy-To-Trigger Poisoning Attack On Code Completion Models: Injecting Disguised Vulnerabilities Against Strong Detection, Shenao Yan, Shen Wang, Yue Duan, Hanbin Hong, Kiho Lee, Doowon Kim, Yuan Hong

Research Collection School Of Computing and Information Systems

Large Language Models (LLMs) have transformed code completion tasks, providing context-based suggestions to boost developer productivity in software engineering. As users often fine-tune these models for specific applications, poisoning and backdoor attacks can covertly alter the model outputs. To address this critical security challenge, we introduce CODEBREAKER, a pioneering LLM-assisted backdoor attack framework on code completion models. Unlike recent attacks that embed malicious payloads in detectable or irrelevant sections of the code (e.g., comments), CODEBREAKER leverages LLMs (e.g., GPT-4) for sophisticated payload transformation (without affecting functionalities), ensuring that both the poisoned data for fine-tuning and generated code can evade strong …

Nonfactoid Question Answering As Query-Focused Summarization With Graph-Enhanced Multihop Inference, Yang Deng, Wenxuan Zhang, Weiwen Xu, Ying Shen, Wai Lam

Research Collection School Of Computing and Information Systems

Nonfactoid question answering (QA) is one of the most extensive yet challenging applications and research areas in natural language processing (NLP). Existing methods fall short of handling the long-distance and complex semantic relations between the question and the document sentences. In this work, we propose a novel query-focused summarization method, namely a graph-enhanced multihop query-focused summarizer (GMQS), to tackle the nonfactoid QA problem. Specifically, we leverage graph-enhanced reasoning techniques to elaborate the multihop inference process in nonfactoid QA. Three types of graphs with different semantic relations, namely semantic relevance, topical coherence, and coreference linking, are constructed for explicitly capturing the …

Style: Improving Domain Transferability Of Asking Clarification Questions In Large Language Model Powered Conversational Agents, Yue Chen, Chen Huang, Yang Deng, Wenqiang Lei, Dingnan Jin, Jia Liu, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Equipping a conversational search engine with strategies regarding when to ask clarification questions is becoming increasingly important across various domains. Attributing to the context understanding capability of LLMs and their access to domain-specific sources of knowledge, LLM-based clarification strategies feature rapid transfer to various domains in a posthoc manner. However, they still struggle to deliver promising performance on unseen domains, struggling to achieve effective domain transferability. We take the first step to investigate this issue and existing methods tend to produce one-size-fits-all strategies across diverse domains, limiting their search effectiveness. In response, we introduce a novel method, called STYLE, to …

On The Multi-Turn Instruction Following For Conversational Web Agents, Yang Deng, Xuan Zhang, Wenxuan Zhang, Yifei Yuan, See-Kiong Ng, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Web agents powered by Large Language Models (LLMs) have demonstrated remarkable abilities in planning and executing multi-step interactions within complex web-based environments, fulfilling a wide range of web navigation tasks. Despite these advancements, the potential for LLM-powered agents to effectively engage with sequential user instructions in real-world scenarios has not been fully explored. In this work, we introduce a new task of Conversational Web Navigation, which necessitates sophisticated interactions that span multiple turns with both the users and the environment, supported by a specially developed dataset named Multi-Turn Mind2Web (MT-Mind2Web). To tackle the limited context length of LLMs and the …

Watme: Towards Lossless Watermarking Through Lexical Redundancy, Liang Chen, Yatao Bian, Yang Deng, Deng Cai, Shuaiyi Li, Peilin Zhao, Kam-Fai Wong

Research Collection School Of Computing and Information Systems

Text watermarking has emerged as a pivotal technique for identifying machine-generated text. However, existing methods often rely on arbitrary vocabulary partitioning during decoding to embed watermarks, which compromises the availability of suitable tokens and significantly degrades the quality of responses. This study assesses the impact of watermarking on different capabilities of large language models (LLMs) from a cognitive science lens. Our finding highlights a significant disparity; knowledge recall and logical reasoning are more adversely affected than language generation. These results suggest a more profound effect of watermarking on LLMs than previously understood. To address these challenges, we introduce Watermarking with …

Prompt Tuning On Graph-Augmented Low-Resource Text Classification, Zhihao Wen, Yuan Fang

Research Collection School Of Computing and Information Systems

Text classification is a fundamental problem in information retrieval with many real-world applications, such as predicting the topics of online articles and the categories of e-commerce product descriptions. However, low-resource text classification, with no or few labeled samples, presents a serious concern for supervised learning. Meanwhile, many text data are inherently grounded on a network structure, such as a hyperlink/citation network for online articles, and a user-item purchase network for e-commerce products. These graph structures capture rich semantic relationships, which can potentially augment low-resource text classification. In this paper, we propose a novel model called Graph-Grounded Pre-training and Prompting (G2P2) …

Macrohft : Memory Augmented Context-Aware Reinforcement Learning On High Frequency Trading, Chuqiao Zong, Chaojie Wang, Molei Qin, Lei Feng, Xinrun Wang, Xinrun Wang

Research Collection School Of Computing and Information Systems

High-frequency trading (HFT) that executes algorithmic trading in short time scales, has recently occupied the majority of cryptocurrency market. Besides traditional quantitative trading methods, reinforcement learning (RL) has become another appealing approach for HFT due to its terrific ability of handling high-dimensional financial data and solving sophisticated sequential decision-making problems, e.g., hierarchical reinforcement learning (HRL) has shown its promising performance on second-level HFT by training a router to select only one sub-agent from the agent pool to execute the current transaction. However, existing RL methods for HFT still have some defects: 1) standard RL-based trading agents suffer from the overfitting …

Reinforcement Tuning For Detecting Stances And Debunking Rumors Jointly With Large Language Models, Ruichao Yang, Wei Gao, Jing Ma, Hongzhan Ling, Bo Wang

Research Collection School Of Computing and Information Systems

Learning multi-task models for jointly detecting stance and verifying rumors poses challenges due to the need for training data of stance at post level and rumor veracity at claim level, which are difficult to obtain. To address this issue, we leverage large language models (LLMs) as the foundation annotators for the joint stance detection (SD) and rumor verification (RV) tasks, dubbed as JSDRV. We introduce a novel reinforcement tuning framework to enhance the joint predictive capabilities of LLM-based SD and RV components. Specifically, we devise a policy for selecting LLM-annotated data at the two levels, employing a hybrid reward mechanism …

Fuel-Saving Route Planning With Data-Driven And Learning-Based Approaches: A Systematic Solution For Harbor Tugs, Shengming Wang, Xiaocai Zhang, Jing Li, Xiaoyang Wei, Hoong Chuin Lau, Bing Tian Dai, Binbin Huang Huang, Zhe Xiao, Xiuju Fu, Zheng Qin

Research Collection School Of Computing and Information Systems

In recent years, there are trends toward cleaner port environments through enforcement by imposed legislation. Transit optimisation of fuel-based port service boats like harbour tugs has emerged as a critical task to reduce fuel consumption and carbon emission. In this paper, an innovative learning-based method, comprising a Reinforcement Learning (RL) model together with a fuel consumption prediction model, was proposed to formulate fuel-saving transit routes. Firstly, an ensemble model is established by combining a Long Short-Term Memory (LSTM) model with a Multilayer Perceptron (MLP) model, predicting fuel use based on tugboat movement and environment factors. Subsequently, an innovative RL based …

Anopay: Anonymous Payment For Vehicle Parking With Updatable Credential, Yang Yang, Wenyi Xue, Yonghua Zhan, Minming Huang, Yingjiu Li, Robert H. Deng

Research Collection School Of Computing and Information Systems

Many existing anonymous parking payment schemes lack high efficiency and flexibility. For instance, the calculation and communication costs involved in payment may linearly increase with the payment amount. In this paper, we propose an anonymous payment system (dubbed AnoPay) for vehicle parking, which leverages updatable attribute-based anonymous credentials and efficient zero-knowledge proof (ZKP) to achieve user anonymity and constant overhead for parking fee payment. To further improve the efficiency, we design a secure parking fee aggregation protocol based on linear homomorphic encryption to aggregate parking transactions, where the amount of each parking transaction is hidden and the privacy of the …

Is Aggregation The Only Choice? Federated Learning Via Layer-Wise Model Recombination, Ming Hu, Zhihao Yue, Xiaofei Xie, Cheng Chen Chen

Research Collection School Of Computing and Information Systems

Although Federated Learning (FL) enables global model training Xiaofei Xie [email protected] Singapore Management University Singapore, Singapore Xian Wei [email protected] East China Normal University Shanghai, China Mingsong Chen∗ [email protected] East China Normal University Shanghai, China • Computing methodologies → Distributed artificial intelligence. across clients without compromising their raw data, due to the unevenly distributed data among clients, existing Federated Averaging (FedAvg)-based methods suffer from the problem of low inference performance. Specifically, different data distributions among clients lead to various optimization directions of local models. Aggregating local models usually results in a low-generalized global model, which performs worse on most of the …

Analyzing Temporal Complex Events With Large Language Models? A Benchmark Towards Temporal, Long Context Understanding, Zhihan Zhang, Yixin Cao, Chenchen Ye, Ma. Yunshan, Lizi Liao, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

The digital landscape is rapidly evolving with an ever-increasing volume of online news, emphasizing the need for swift and precise analysis of complex events.We refer to the complex events composed of many news articles over an extended period as Temporal Complex Event (TCE). This paper proposes a novel approach using Large Language Models (LLMs) to systematically extract and analyze the event chain within TCE, characterized by their key points and timestamps. We establish a benchmark, named TCELongBench, to evaluate the proficiency of LLMs in handling temporal dynamics and understanding extensive text. This benchmark encompasses three distinct tasks - reading comprehension, …

"The Hotel Of 10,000 Stars": The Impact Of Social-Structural Determinants Of Health Among Im/Migrant Shrimpers In The Gulf Of Mexico, Shannon Guillot-Wright, Lacy Davis, Linh Truong, Heide Castañeda, Anabel Rodriguez

Faculty, Staff and Student Publications

Objectives. To identify appropriate interventions to prevent injury, we conducted a qualitative study among commercial shrimp fishermen in the Gulf of Mexico.

Methods. Using qualitative and participatory research methods, including interviews, photovoice, and workplace observations in southeast Texas and the Rio Grande Valley in Texas, we examined the social‒structural dimensions that contribute to physical and psychological injury.

Results. We found that multiple layers of vulnerability and danger exist among shrimpers with interacting themes: (1) recognizing risk, (2) precarious employment, and (3) psychological distress.

Conclusions. Our results add to the growing body of knowledge that emphasizes the negative health impacts of …

Comprehensive Investigation Into The Molecular Mechanisms Driving The Emergence Of Emm4 Group A Streptococcus, Chioma Odo, Samuel A Shelburne, Anthony R. Flores, Steven J Norris, Anna Konovalova, Blake Hanson

Dissertations and Theses (Open Access)

Abstract

Comprehensive Investigation into the Molecular Mechanisms Driving the Emergence of emm4 Group A Streptococcus

Chioma Modeline Odo, M.S.

Advisory Professor: Samuel A. Shelburne, MD, Ph.D.

The major gram-positive bacterium group A Streptococcus (GAS) is a model organism for studying microbial epidemics as it is well known to cause infections that occur in waves. GAS strains are grouped based on the composition of the N-terminal sequence of the M protein which is encoded by the emm gene. With the advent of large-scale whole genome sequencing, GAS clonal emergence events in emm1, emm3, and emm89 types have been identified …

Consistency Of Advanced Radiation Therapy Qa Methods And Global Auditing Systems, Fre'etta Brooks

Dissertations and Theses (Open Access)

Consistency of Advanced Radiation Therapy QA Methods and Global Auditing Systems

Fre’Etta M.D. Brooks, B.S.

Advisory Professor: Stephen F. Kry, Ph.D.

Abstract

Auditing systems provide a means to verify the implementation and operation of all equipment and radiation parameters used by an institution providing radiation therapy treatments. The systems are important tools that are used to determine if an institution can accurately deliver radiation therapy that requires complex planning in the form of IMRT or VMAT and credential radiotherapy treatment institutions for specific clinical trial protocols; to assure the quality of treatment to all patients. Auditing techniques vary by region …

Delineating Metastasis In Mss Colorectal Cancer By Single - Cell Rna Sequencing, Shuangjie You

Dissertations and Theses (Open Access)

Colorectal cancer Colorectal Cancer (CRC) is the leading cause of cancer-related deaths, with the majority of these deaths occurring after metastases. In this study, we used single-cell RNA transcriptome sequencing to characterize the cellular and molecular features of primary CRC and colorectal liver metastases (CRLM). Analysis of 130 samples revealed significant heterogeneity in the tumor microenvironment (TME) characterized by a diversity of cancer-associated fibroblasts, immune cells, and endothelial cell populations. Notably, we identified different CAF subtypes, including inflammatory (iCAFs), matrix(mCAFs), complement (cCAFs), and vascular (vCAFs). iCAFs were enriched in primary tumors expressing higher levels of inflammation-associated genes, whereas mCAFs were …

Single Cell And Spatial Analysis Characterize Metabolic Dpep1+ Fibroblasts In Pdac, Yuanyuan Zhu

Dissertations and Theses (Open Access)

Pancreatic ductal adenocarcinoma (PDAC) is highly malignant and exhibits aggressive progression patterns, resists conventional treatments, and poses significant challenges to patient survival, necessitating urgent research and development of novel therapeutic approaches. Understanding cancer-associated fibroblasts (CAF) metabolism reprogramming in PDAC tumor microenvironment (TME) may uncover novel therapeutic targets. Here, we used single cell and spatial transcriptomics to comprehensively characterize the TME content of human PDAC. We characterized CAFs metabotypes and a novel CAF population expressing DPEP1, COMP, CST1, which involved in glutathione metabolism, termed "gluCAFs” and was associated with poor clinical outcomes and advanced PDAC stages. We also validate the existence …

Tsyn-Seq: A T Cell Synapse-Based Antigen Identification Platform, Yimei Jin

Dissertations and Theses (Open Access)

Tools for genome-wide rapid identification of peptide–major histocompatibility complex targets of T-cell receptors (TCRs) are not yet universally available. We present a new antigen screening method, the T-synapse (Tsyn) reporter system, which includes antigen-presenting cells (APCs) with a Fas-inducible NF-κB reporter and T cells with a nuclear factor of activated T cells (NFAT) reporter. To functionally screen for target antigens from a cDNA library, productively interacting T cell–APC aggregates were detected by dual reporter activity and enriched by flow sorting followed by antigen identification quantified by deep sequencing (Tsyn-seq). When applied to a previously characterized TCR specific for the E7 …

Exploring The Role Of The Arginine-Methylation Writer-Reader Pair Prmt5/Snd1 In Jak2-Mutant Myeloproliferative Neoplasms, Rocio Rubiano

Dissertations and Theses (Open Access)

Myeloproliferative neoplasms (MPNs) are a hematopoietic disease characterized by hyperproliferation of cells of the myeloid lineage for which current therapeutic options are limited. Discovered in 2005, the JAK2V617F mutation is the most common driver mutation in BCR-ABL negative MPNs, resulting in constitutive activation of the JAK2 protein and the JAK-STAT signaling pathway. A role for the methyltransferase activity of Protein Arginine Methyltransferase 5 (PRMT5) has been proposed in JAK2-mutant MPN, highlighting both a mechanism through which this mutation can drive disease progression and a potential mode of therapeutic intervention. Staphylococcal Nuclease Domain-Containing Protein 1 (SND1) is the effector molecule responsible …

Influential Factors And Predicting Dose Delivery Accuracy For Imaging And Radiation Oncology Core’S Phantom Program Using Machine Learning, Hunter Mehrens

Dissertations and Theses (Open Access)

IROC’s mission is to help ensure consistent and comparable, high-quality radiotherapy across clinics that participate in national clinical trials. To obtain this mission, IROC’s phantom program provides a third-party end-to-end check of the clinical workflow of a patient receiving radiotherapy. The goal of the phantom audit is to compare the dose delivered to the dose planned by the treatment system ensuring dose delivery accuracy. While IROC’s phantoms are better equipped to catch dose delivery errors compared to a clinic’s QA process, the end-to-end process and reporting of results is time-consuming creating a bottleneck for clinical trial participation. Furthermore, IROC’s passing …

The Role Of An Ultraconserved Long Non-Coding Rna In B-Cell Lymphomagenesis, Swati Mohapatra

Dissertations and Theses (Open Access)

Ultraconserved regions (UCRs) are genomic segments with perfect (100%) conservation between the orthologous regions of human, rat, and mouse genomes. UCRs can be transcribed into mono-exonic long non-coding RNAs (lncRNAs) known as transcribed ultraconserved regions (T-UCRs). These regions, despite lacking protein-coding potential, are increasingly recognized for their regulatory roles in gene expression, including the modulation of non-coding RNA (ncRNA) transcripts. NcRNAs play crucial roles in cellular processes, including oncogenic transformation, with emerging evidence revealing their ability to encode small peptides known as ncRNA-encoded peptides (ncPEPs). These peptides, originating from small open reading frames (smORFs), contribute to diverse cellular functions and …

Evaluation Of Myogenic Differentiation Potential Of Bmd Ipscs In Vitro And The Effect Of Ubiquitination Inhibitors On Dystrophin Stability In Vivo, Muchen Liu

Dissertations and Theses (Open Access)

Background: Muscular dystrophies are heterogeneous groups of inherited diseases leading to progressive muscular weakness and degeneration. In the case of Becker muscular dystrophy (BMD), non-disrupting mutations of the DMD gene reading frame is the causative defect. By the age of 30s, about 60 to 70% of BMD patients develop cardiomyopathy, which is often lethal. Unfortunately, there is currently no cure for BMD. Meanwhile, improvement of dystrophin stability by repressing polyubiquitination is a promising strategy. The application of human iPSCs in tissue regeneration also serves as a potential therapeutic strategy. This study aims to test the efficacy of selected drugs interfering …

The Effect Of Context Diversity In Virtual Reality Environments On Speech Motor Learning In A Person With Aphasia And Apraxia Of Speech, Shelby Keeler Holloway

Theses and Dissertations

This study explored the effect of contextual diversity on speech motor learning in an individual with apraxia of speech (AOS) using virtual reality (VR). One 62-year-old participant with severe AOS participated in the study where he practiced monosyllabic CVC words in both consistent and diverse VR environments. Kinematic, acoustic, and perceptual data were collected and evaluated for evidence of learning. Kinematic data were collected using electromagnetic articulography where sensors were glued to the participant's upper lip, lower lip, tongue blade, and tongue tip and were tracked with high precision. In addition, acoustic and perceptual data were collected using a microphone …

How Fun Overcame Fear: The Gamification Of A Graduate-Level Statistics Course, Mai P. Trinh, Robert J. Chico, Rachel M. Reed

Management Faculty Publications

Innovative instructional methods can help improve student engagement and learning outcomes when teaching difficult subjects, such as statistics. This instructional innovation article illustrates how gamification can be applied in management education to improve students’ learning experience, engagement, and acquisition of knowledge. Our purpose is to demonstrate how gamification is not only a powerful way to build on the use of games and game thinking in our field, but also a versatile application of education technology that could potentially enhance the way management knowledge is taught. Furthermore, it is a low-risk way for management educators to join and contribute to the …

Large-Scale Hpc-Empowered Power Electronics Modeling And Simulation In Photovoltaic Applications, Liwei Wang

All Dissertations

The rising popularity of renewable energy sources requires advanced, efficient power electronic systems for energy conversion, grid integration, and system management, thereby raising expectations for power electronics in the energy industry. The complexity of modern power electronic systems requires comprehensive simulations and in-depth analysis to predict performance accurately, but this process is impeded by prolonged simulation times. The primary objective of this dissertation is to develop a high-fidelity, high-speed event-driven simulator to tackle challenges related to mass data processing, uncertainty evaluation, as well as modeling and simulation issues in assessing the reliability of power electronics in large-scale Photovoltaic (PV) systems. …

Time-Domain Line Protection In Presence Of Renewables, Prabin Adhikari

All Dissertations

Inverter based resources (IBRs) are crucial in integrating renewable energy sources into the power grid. However, their unique fault characteristics, significantly different from synchronous generators (SGs), present several challenges for existing line protection schemes at both transmission and distribution levels. These schemes, reliant on distance and directional relays designed in phasor domain, are not well-suited for IBRs. Most published literature addressing this problem concentrates on altering the control design of inverters. However, this approach faces practical limitations. Inverter controls, often proprietary, are not readily accessible to utilities, rendering the control-based solutions impractical for widespread implementation.

To address this challenge, this …

Parent Advocacy For Literacy Intervention: A Case Study Of Motivations And Experiences, Meghan Malloy

All Dissertations

While systems of support are used in schools to assist students’ academic growth, parents are not always informed of the support process or the academic goals set for their child (RTI Action Network, n.d.; Troisi, 2014; Weingarten et al., 2020). In the aftermath of the pandemic, students present a wide variety of academic needs (Lewis et al., 2021; Maldonado & De Witte, 2020), underscoring the importance of parent advocacy. In situations where a student needs additional supports, such as literacy intervention, the parent may become the child’s best advocate (Besnoy et al., 2015).

Using a multiple case study design, the …

Fluid-Solid Coupled Analysis Of Biological Samples In A Flow Chamber Using Optical Coherence Tomography, Reece W. Fratus

All Dissertations

Forces from fluid flow on a tissue or cellular boundary can drive remodeling processes through mechanotransduction pathways. In most cases, the exact mechanisms are not understood, such as in marine biofilm development or vascular remodeling. To gain a better understanding of these interactions, the flow profile at a solid boundary and the mechanical changes of that solid can be coupled together to determine the impact of fluid flow on mechanical remodeling. To achieve this, optical coherence tomography (OCT) is used to image the fluid and solid simultaneously. Fluid seeded with particles can be imaged and run through a particle image …