Physical Sciences and Mathematics Commons^™

Pollutant Forecasting Using Neural Network-Based Temporal Models, Richard Pike

Masters Theses & Specialist Projects

The Jing-Jin-Ji region of China is a highly industrialized and populated area of the country. Its periodic high pollution and smog includes particles smaller than 2.5 μm, known as PM2.5, linked to many respiratory and cardiovascular illnesses. PM2.5 concentration around Jing-Jin-Ji has exceeded China’s urban air quality safety threshold for over 20% of all days in 2017 through 2020.

The quantity of ground weather stations that measure the concentrations of these pollutants, and their valuable data, is unfortunately small. By employing many machine learning strategies, many researchers have focused on interpolating finer spatial grids of PM2.5, or hindcasting PM2.5. However, …

Llm-Adapters: An Adapter Family For Parameter-Efficient Fine-Tuning Of Large Language Models, Zhiqiang Hu, Lei Wang, Yihuai Lan, Wanyu Xu, Ee-Peng Lim, Lidong Bing, Xing Xu, Soujanya Poria, Roy Ka-Wei Lee

Research Collection School Of Computing and Information Systems

The success of large language models (LLMs), like GPT-4 and ChatGPT, has led to the development of numerous cost-effective and accessible alternatives that are created by finetuning open-access LLMs with task-specific data (e.g., ChatDoctor) or instruction data (e.g., Alpaca). Among the various fine-tuning methods, adapter-based parameter-efficient fine-tuning (PEFT) is undoubtedly one of the most attractive topics, as it only requires fine-tuning a few external parameters instead of the entire LLMs while achieving comparable or even better performance. To enable further research on PEFT methods of LLMs, this paper presents LLMAdapters, an easy-to-use framework that integrates various adapters into LLMs and …

Benchmarking Foundation Models With Language-Model-As-An-Examiner, Yushi Bai, Jiahao Ying, Yixin Cao, Xin Lv, Yuze He, Xiaozhi Wang, Jifan Yu, Kaisheng Zeng, Yijia Xiao, Haozhe Lyu, Jiayin Zhang, Juanzi Li, Lei Hou

Research Collection School Of Computing and Information Systems

Numerous benchmarks have been established to assess the performance of foundation models on open-ended question answering, which serves as a comprehensive test of a model’s ability to understand and generate language in a manner similar to humans. Most of these works focus on proposing new datasets, however, we see two main issues within previous benchmarking pipelines, namely testing leakage and evaluation automation. In this paper, we propose a novel benchmarking framework, Language-Model-as-an-Examiner, where the LM serves as a knowledgeable examiner that formulates questions based on its knowledge and evaluates responses in a reference-free manner. Our framework allows for effortless extensibility …

The Psychological Science Accelerator's Covid-19 Rapid-Response Dataset, Erin M. Buchanan, Andree Hartanto

Research Collection School of Social Sciences

In response to the COVID-19 pandemic, the Psychological Science Accelerator coordinated three large-scale psychological studies to examine the effects of loss-gain framing, cognitive reappraisals, and autonomy framing manipulations on behavioral intentions and affective measures. The data collected (April to October 2020) included specific measures for each experimental study, a general questionnaire examining health prevention behaviors and COVID-19 experience, geographical and cultural context characterization, and demographic information for each participant. Each participant started the study with the same general questions and then was randomized to complete either one longer experiment or two shorter experiments. Data were provided by 73,223 participants with …

Leveraging Generative Agents: Autonomous Ai With Simulated Personas For Interactive Simulacra And Collaborative Research, James Hutson, Jay Ratican

Faculty Scholarship

The advent of large language models (LLMs) and AI learning have fundamentally reshaped the research landscape, paving the way for novel problem-solving approaches. This paper introduces a unique framework that leverages the capabilities of autonomous AI agents with simulated personas to drive collaborative research in groundbreaking ways. Inspired by a recent study of autonomous agents mirroring human behavior, this concept encourages the use of a cadre of AI agents, each possessing specialized expertise for collective endeavors. By replicating human diversity in teamwork, this approach targets complex and hitherto unsolvable issues. The key to this strategy is persona and emotional simulation, …

Globally-Distributed Microbial Eukaryotes Exhibit Endemism At Deep-Sea Hydrothermal Vents, Sarah K. Hu, Amy R. Smith, Rika E. Anderson, Sean P. Sylva, Michael Setzer, Maria Steadmon, Kiana L. Frank, Eric W. Chan, Darlene S. S. Lim, John A. Breier

School of Earth, Environmental, and Marine Sciences Faculty Publications and Presentations

Single-celled microbial eukaryotes inhabit deep-sea hydrothermal vent environments and play critical ecological roles in the vent-associated microbial food web. 18S rRNA amplicon sequencing of diffuse venting fluids from four geographically- and geochemically-distinct hydrothermal vent fields was applied to investigate community diversity patterns among protistan assemblages. The four vent fields include Axial Seamount at the Juan de Fuca Ridge, Sea Cliff and Apollo at the Gorda Ridge, all in the NE Pacific Ocean, and Piccard and Von Damm at the Mid-Cayman Rise in the Caribbean Sea. We describe species diversity patterns with respect to hydrothermal vent field and sample type, identify …

Qasm-To-Hls: A Framework For Accelerating Quantum Circuit Emulation On High-Performance Reconfigurable Computers, Anshul Maurya

Theses and Dissertations

High-performance reconfigurable computers (HPRCs) make use of Field-Programmable Gate Arrays (FPGAs) for efficient emulation of quantum algorithms. Generally, algorithm-specific architectures are implemented on the FPGAs and there is very little flexibility. Moreover, mapping a quantum algorithm onto its equivalent FPGA emulation architecture is challenging. In this work, we present an automation framework for converting quantum circuits to their equivalent FPGA emulation architectures. The framework processes quantum circuits represented in Quantum Assembly Language (QASM) and derives high-level descriptions of the hardware emulation architectures for High-Level Synthesis (HLS) on HPRCs. The framework generates the code for a heterogeneous architecture consisting of a …

Zero-Knowledge Reductions And Confidential Arithmetic, Marvin Jones

All Dissertations

The changes in computing paradigms to shift computations to third parties have resulted in the necessity of these computations to be provable. Zero-knowledge arguments are probabilistic arguments that are used to to verify computations without secret data being leaked to the verifying party.

In this dissertation, we study zero-knowledge arguments with specific focus on reductions. Our main contributions are:

1. Provide a thorough survey in a variety of zero-knowledge techniques and protocols.
2. Prove various results of reductions that can be used to study interactive protocols in terms of subroutines. Additionally, we identify an issue in the analogous definition of zero-knowledge for …

Studies On Electrochemical Hydrogen Isotope Separation, Liyanage Mayura Sankalpa Silva

All Dissertations

Graphene-integrated Proton Exchange Membrane (PEM) electrochemical cells have emerged as a novel area of scientific investigation in the realm of hydrogen isotope separation. Chemical Vapor Deposited (CVD) graphene has been especially useful due to its large-scale production capability for scaling-up purposes. The research described in this dissertation explores the role that inadvertent introduction of cations, notably ammonium and copper, during the CVD graphene transfer onto PEM substrates, such as Nafion, might play in affecting hydrogen ion transport and isotope separation in PEM electrochemical cells. An extensive review of existing literature exposed a gap concerning unintentional cation introductions during graphene transfer, …

Leveraging Artificial Intelligence For Team Cognition In Human-Ai Teams, Beau Schelble

All Dissertations

Advances in artificial intelligence (AI) technologies have enabled AI to be applied across a wide variety of new fields like cryptography, art, and data analysis. Several of these fields are social in nature, including decision-making and teaming, which introduces a new set of challenges for AI research. While each of these fields has its unique challenges, the area of human-AI teaming is beset with many that center around the expectations and abilities of AI teammates. One such challenge is understanding team cognition in these human-AI teams and AI teammates' ability to contribute towards, support, and encourage it. Team cognition is …

Atomistic And Mesoscale Modeling Of Microstructure Development During Solid-State Sintering, Omar Marwan Isa Hussein

All Dissertations

Interfaces are ubiquitous in materials systems, and they influence the processing and properties of nearly all engineering and functional materials. Examples include grain boundaries (GBs) in polycrystalline materials, free surfaces in nanoparticles, and phase boundaries in multiphase materials. Therefore, understanding and controlling interfacial processes is a key aspect of materials design and discovery efforts. Recent developments in advanced manufacturing and synthesis techniques have enabled the fabrication of materials architectures with intricate nanoscale features. Of particular interest is solid-state sintering, known for creating complex and high-precision geometries with controlled microstructures. While sintering science has been the subject of active research, very …

New Preconditioned Conjugate Gradient Methods For Some Structured Problems In Physics, Tianqi Zhang

All Dissertations

This dissertation concerns the development and analysis of new preconditioned conjugate gradient (PCG) algorithms for three important classes of large-scale and complex physical problems characterized by special structures. We propose several new iterative methods for solving the eigenvalue problem or energy minimization problem, which leverage the unique structures inherent in these problems while preserving the underlying physical properties. The new algorithms enable more efficient and robust large-scale modeling and simulations in many areas, including condensed matter physics, optical properties of materials, stabilities of dynamical systems arising from control problems, and many more. Some methods are expected to be applicable to …

Fledge: Ledger-Based Federated Learning Resilient To Inference And Backdoor Attacks, Jorge Castillo, Phillip Rieger, Hossein Fereidooni, Qian Chen, Ahmad Sadeghi

Informatics and Engineering Systems Faculty Publications and Presentations

Federated learning (FL) is a distributed learning process that uses a trusted aggregation server to allow multiple parties (or clients) to collaboratively train a machine learning model without having them share their private data. Recent research, however, has demonstrated the effectiveness of inference and poisoning attacks on FL. Mitigating both attacks simultaneously is very challenging. State-of-the-art solutions have proposed the use of poisoning defenses with Secure Multi-Party Computation (SMPC) and/or Differential Privacy (DP). However, these techniques are not efficient and fail to address the malicious intent behind the attacks, i.e., adversaries (curious servers and/or compromised clients) seek to exploit a …

Aspects Of Stochastic Geometric Mechanics In Molecular Biophysics, David Frost

All Dissertations

In confocal single-molecule FRET experiments, the joint distribution of FRET efficiency and donor lifetime distribution can reveal underlying molecular conformational dynamics via deviation from their theoretical Forster relationship. This shift is referred to as a dynamic shift. In this study, we investigate the influence of the free energy landscape in protein conformational dynamics on the dynamic shift by simulation of the associated continuum reaction coordinate Langevin dynamics, yielding a deeper understanding of the dynamic and structural information in the joint FRET efficiency and donor lifetime distribution. We develop novel Langevin models for the dye linker dynamics, including rotational dynamics, based …

Gated Recurrent Units For Blockage Mitigation In Mmwave Wireless, Ahmed H. Almutairi, Alireza Keshavarz-Haddad, Ehsan Aryafar

Computer Science Faculty Publications and Presentations

Millimeter-Wave (mmWave) communication is susceptible to blockages, which can significantly reduce the signal strength at the receiver. Mitigating the negative impacts of blockages is a key requirement to ensure reliable and high throughput mmWave communication links. Previous research on blockage mitigation has introduced several model and protocol based blockage mitigation solutions that focus on one technique at a time, such as handoff to a different base station or beam adaptation to the same base station. In this paper, we address the overarching problem: what blockage mitigation method should be employed? and what is the optimal sub-selection within that method? To …

Preventing Inferences Through Data Dependencies On Sensitive Data, Primal Pappachan, Shufan Zhang, Xi He, Sharad Mehrotra

Computer Science Faculty Publications and Presentations

Simply restricting the computation to non-sensitive part of the data may lead to inferences on sensitive data through data dependencies. Inference control from data dependencies has been studied in the prior work. However, existing solutions either detect and deny queries which may lead to leakage – resulting in poor utility, or only protects against exact reconstruction of the sensitive data – resulting in poor security. In this paper, we present a novel security model called full deniability. Under this stronger security model, any information inferred about sensitive data from non-sensitive data is considered as a leakage. We describe algorithms for …

The Effects Of Salt Marsh Restoration On The Hydrology Of Salt Marsh Channels, Isis Kontas

University Honors Theses

Salt marshes produce many ecosystem services, from water purification to protection from hurricanes. Despite their benefits, salt marshes have been impacted negatively by human activities. There are many salt marsh restoration projects that intend to bring back all ecological functions and services. Quantifiable measurements are needed to evaluate the effectiveness of such restoration efforts. Earlier work by Reagan Thomas demonstrated what happens to the hydrology of salt marsh channels when they are adjacent to restored salt marshes. This study builds on Thomas’ work and uses the sinuosity of channels as a quantitative, representative metric of salt marsh hydrology restoration effectiveness. …

Parameterized Complexity Of Feature Selection For Categorical Data Clustering, Sayan Bandyapadhyay, Fedor V. Fomin, Petr A. Golovach, Kirill Simonov

Computer Science Faculty Publications and Presentations

We develop new algorithmic methods with provable guarantees for feature selection in regard to categorical data clustering. While feature selection is one of the most common approaches to reduce dimensionality in practice, most of the known feature selection methods are heuristics. We study the following mathematical model. We assume that there are some inadvertent (or undesirable) features of the input data that unnecessarily increase the cost of clustering. Consequently, we want to select a subset of the original features from the data such that there is a small-cost clustering on the selected features. More precisely, for given integers ℓ (the …

Coral Restoration: Comparisons In Space, Time, Impacts, And Costs, Allison Fargo

Honors College

Seventy-five percent of coral reefs globally face crisis due to anthropogenic disturbances, prompting heightened global coral restoration initiatives to preserve these vital ecosystems. Various regions employ diverse active coral restoration methodologies, including coral gardening, transplantation, micro-fragmentation, artificial reefs, and sexual propagation. Of these methods, coral gardening stands out as one of the most common and highly successful methods, alongside widespread transplantation practices. Restoration efforts predominantly focus on acroporids due to their relatively rapid growth and asexual fragmentation; however, a diverse range of coral species, including large, slow-growing varieties, is also employed in these endeavors. Costs vary significantly, ranging from $10,000 …

Neural Multi-Objective Combinatorial Optimization With Diversity Enhancement, Jinbiao Chen, Zizhen Zhang, Zhiguang Cao, Yaoxin Wu, Yining Ma, Te Ye, Jiahai Wang

Research Collection School Of Computing and Information Systems

Most of existing neural methods for multi-objective combinatorial optimization (MOCO) problems solely rely on decomposition, which often leads to repetitive solutions for the respective subproblems, thus a limited Pareto set. Beyond decomposition, we propose a novel neural heuristic with diversity enhancement (NHDE) to produce more Pareto solutions from two perspectives. On the one hand, to hinder duplicated solutions for different subproblems, we propose an indicator-enhanced deep reinforcement learning method to guide the model, and design a heterogeneous graph attention mechanism to capture the relations between the instance graph and the Pareto front graph. On the other hand, to excavate more …

Designing An Overseas Experiential Course In Data Science, Hua Leong Fwa, Graham Ng

Research Collection School Of Computing and Information Systems

Unprecedented demand for data science professionals in the industry has led to many educational institutions launching new data science courses. It is however imperative that students of data science programmes learn through execution of real-world, authentic projects on top of acquiring foundational knowledge on the basics of data science. In the process of working on authentic, real-world projects, students not only create new knowledge but also learn to solve open, sophisticated, and ill-structured problems in an inter-disciplinary fashion. In this paper, we detailed our approach to design a data science curriculum premised on learners solving authentic data science problems sourced …

Learning To Search Feasible And Infeasible Regions Of Routing Problems With Flexible Neural K-Opt, Yining Ma, Zhiguang Cao, Yew Meng Chee

Research Collection School Of Computing and Information Systems

In this paper, we present Neural k-Opt (NeuOpt), a novel learning-to-search (L2S) solver for routing problems. It learns to perform flexible k-opt exchanges based on a tailored action factorization method and a customized recurrent dual-stream decoder. As a pioneering work to circumvent the pure feasibility masking scheme and enable the autonomous exploration of both feasible and infeasible regions, we then propose the Guided Infeasible Region Exploration (GIRE) scheme, which supplements the NeuOpt policy network with feasibility-related features and leverages reward shaping to steer reinforcement learning more effectively. Besides, we further equip NeuOpt with dynamic data augmentations during inference for more …

Deepaco: Neural-Enhanced Ant Systems For Combinatorial Optimization, Haoran Ye, Jiarui Wang, Zhiguang Cao, Helan Liang, Yong Li

Research Collection School Of Computing and Information Systems

Ant Colony Optimization (ACO) is a meta-heuristic algorithm that has been successfully applied to various Combinatorial Optimization Problems (COPs). Traditionally, customizing ACO for a specific problem requires the expert design of knowledge-driven heuristics. In this paper, we propose DeepACO, a generic framework leveraging deep reinforcement learning to automate heuristic designs. DeepACO serves to strengthen the heuristic measures of existing ACO algorithms and dispense with laborious manual design in future ACO applications. As a neural-enhanced meta-heuristic, DeepACO consistently outperforms its ACO counterparts on eight COPs using a single neural model and a single set of hyperparameters. As a Neural Combinatorial Optimization …

Efficient Meta Neural Heuristic For Multi-Objective Combinatorial Optimization, Jinbiao Chen, Zizhen Zhang, Te Ye, Zhiguang Cao, Siyuan Chen, Jiahai Wang

Research Collection School Of Computing and Information Systems

Recently, neural heuristics based on deep reinforcement learning have exhibited promise in solving multi-objective combinatorial optimization problems (MOCOPs). However, they are still struggling to achieve high learning efficiency and solution quality. To tackle this issue, we propose an efficient meta neural heuristic (EMNH), in which a meta model is first trained and then fine-tuned with a few steps to solve corresponding single-objective subproblems. Specifically, for the training process, a (partial) architecture-shared multi-task model is leveraged to achieve parallel learning for the meta model, so as to speed up the training; meanwhile, a scaled symmetric sampling method with respect to the …

Metabox: A Benchmark Platform For Meta-Black-Box Optimization With Reinforcement Learning, Zeyuan Ma, Hongshu Guo, Jiacheng Chen, Zhenrui Li, Guojun Peng, Yue-Jiao Gong, Yining Ma, Zhiguang Cao

Research Collection School Of Computing and Information Systems

Recently, Meta-Black-Box Optimization with Reinforcement Learning (MetaBBO-RL) has showcased the power of leveraging RL at the meta-level to mitigate manual fine-tuning of lower-level black-box optimizers. However, this field is hindered by the lack of a unified benchmark. To fill this gap, we introduce MetaBox, the first benchmark platform expressly tailored for developing and evaluating MetaBBO-RL methods. MetaBox offers a flexible algorithmic template that allows users to effortlessly implement their unique designs within the platform. Moreover, it provides a broad spectrum of over 300 problem instances, collected from synthetic to realistic scenarios, and an extensive library of 19 baseline methods, including …

Truncated Affinity Maximization: One-Class Homophily Modeling For Graph Anomaly Detection, Hezhe Qiao, Guansong Pang

Research Collection School Of Computing and Information Systems

We reveal a one-class homophily phenomenon, which is one prevalent property we find empirically in real-world graph anomaly detection (GAD) datasets, i.e., normal nodes tend to have strong connection/affinity with each other, while the homophily in abnormal nodes is significantly weaker than normal nodes. However, this anomaly-discriminative property is ignored by existing GAD methods that are typically built using a conventional anomaly detection objective, such as data reconstruction. In this work, we explore this property to introduce a novel unsupervised anomaly scoring measure for GAD – local node affinity – that assigns a larger anomaly score to nodes that are …

A Poisson-Based Distribution Learning Framework For Short-Term Prediction Of Food Delivery Demand Ranges, Jian Liang, Jintao Ke, Hai Wang, Hongbo Ye, Jinjun Tang

Research Collection School Of Computing and Information Systems

The COVID-19 pandemic has caused a dramatic change in the demand composition of restaurants and, at the same time, catalyzed on-demand food delivery (OFD) services—such as DoorDash, Grubhub, and Uber Eats—to a large extent. With massive amounts of data on customers, drivers, and merchants, OFD platforms can achieve higher efficiency with better strategic and operational decisions; these include dynamic pricing, order bundling and dispatching, and driver relocation. Some of these decisions, and especially proactive decisions in real time, rely on accurate and reliable short-term predictions of demand ranges or distributions. In this paper, we develop a Poisson-based distribution prediction (PDP) …

Spatial-Temporal Episodic Memory Modeling For Adls: Encoding, Retrieval, And Prediction, Xinjing Song, Di Wang, Chai Quek, Ah-Hwee Tan, Yanjiang Wang

Research Collection School Of Computing and Information Systems

Activities of daily living (ADLs) relate to people’s daily self-care activities, which reflect their living habits and lifestyle. A prior study presented a neural network model called STADLART for ADL routine learning. In this paper, we propose a cognitive model named Spatial-Temporal Episodic Memory for ADL (STEM-ADL), which extends STADLART to encode event sequences in the form of distributed episodic memory patterns. Specifically, STEM-ADL encodes each ADL and its associated contextual information as an event pattern and encodes all events in a day as an episode pattern. By explicitly encoding the temporal characteristics of events as activity gradient patterns, STEM-ADL …

Exploring Students' Adoption Of Chatgpt As A Mentor For Undergraduate Computing Projects: Pls-Sem Analysis, Gottipati Swapna, Kyong Jin Shim, Shankararaman, Venky

Research Collection School Of Computing and Information Systems

As computing projects increasingly become a core component of undergraduate courses, effective mentorship is crucial for supporting students' learning and development. Our study examines the adoption of ChatGPT as a mentor for undergraduate computing projects. It explores the impact of ChatGPT mentorship, specifically, skills development, and mentor responsiveness, i.e., ChatGPT's responsiveness to students' needs and requests. We utilize PLS-SEM to investigate the interrelationships between different factors and develop a model that captures their contribution to the effectiveness of ChatGPT as a mentor. The findings suggest that mentor responsiveness and technical/design support are key factors for the adoption of AI tools …

Offline Rl With Discrete Proxy Representations For Generalizability In Pomdps, Pengjie Gu, Xinyu Cai, Dong Xing, Xinrun Wang, Mengchen Zhao, Bo An

Research Collection School Of Computing and Information Systems

Offline Reinforcement Learning (RL) has demonstrated promising results in various applications by learning policies from previously collected datasets, reducing the need for online exploration and interactions. However, real-world scenarios usually involve partial observability, which brings crucial challenges of the deployment of offline RL methods: i) the policy trained on data with full observability is not robust against the masked observations during execution, and ii) the information of which parts of observations are masked is usually unknown during training. In order to address these challenges, we present Offline RL with DiscrEte pRoxy representations (ORDER), a probabilistic framework which leverages novel state …