Open Access. Powered by Scholars. Published by Universities.®

Digital Commons Network™

Open Access. Powered by Scholars. Published by Universities.®

2025

Discipline
Institution
Keyword
Publication
Publication Type
File Type

Articles 2161 - 2190 of 25641

Full-Text Articles in Entire DC Network

Design And Optimization Of High-Frequency Medium-Voltage Dual Active Bridge Dc/Dc Converter Using Sic Device, Hui Cao Dec 2025

Design And Optimization Of High-Frequency Medium-Voltage Dual Active Bridge Dc/Dc Converter Using Sic Device, Hui Cao

Graduate Theses and Dissertations

High frequency Dual Active Bridge (DAB) converters have become a key enabling technology in medium- and high-power applications such as renewable energy integration, electric vehicle fast charging, and emerging DC power distribution systems. However, when operating at high power and high frequency, DAB converters face critical challenges including limited soft-switching range, uneven semiconductor loss distribution, deadtime-induced resonance, switching-frequency constraints of wide-bandgap (WBG) devices, and transformer DC bias during dynamic transitions. This dissertation addresses these challenges through advancements in modulation strategies and converter architecture. First, an enhanced triple-phase-shift (E-TPS) modulation scheme is proposed to achieve balanced switching and conduction losses across …

Go to article

Programmable Network Approaches To Resilience And Security In Phasor Measurement Unit Networks, Zhiyao He Dec 2025

Programmable Network Approaches To Resilience And Security In Phasor Measurement Unit Networks, Zhiyao He

Graduate Theses and Dissertations

The security and resilience of smart grids are critical for ensuring reliable and stable power delivery. As modern power systems evolve to incorporate more advanced sensing and control capabilities, Phasor Measurement Units (PMUs) have become an important source of high-frequency, time-synchronized measurements that support wide-area monitoring, control, and protection. However, the growing complexity of smart grids and their reliance on real-time communication expose them to a range of cyber threats, including data loss, tampering, and coordinated attacks. This dissertation explores the use of programmable network technologies, particularly P4-based programmable switches, to provide in-network solutions that enhance the reliability and security …

Go to article

Artificial Intelligence For Reliability: Predictive Health Maintenance And Geolocation In Gps-Denied Environments, Rafael Toche Pizano Dec 2025

Artificial Intelligence For Reliability: Predictive Health Maintenance And Geolocation In Gps-Denied Environments, Rafael Toche Pizano

Graduate Theses and Dissertations

In this dissertation, we explore the potential of machine learning and deep learning techniques to enhance the performance and robustness of applications across two major domains. By addressing the challenges within these fields, we demonstrate that we can leverage learning algorithms to obtain substantial improvements in accuracy and robustness. First, we tackle a problem in the field of predictive health maintenance. We propose a novel auto encoder and neural network based methodology to predict failure times in complex aviation systems to learn to distinguish between normal and abnormal operational behavior, and use this information to inform the neural network to …

Go to article

Agricultural Practices’ Impact On Soil Health Indicators In Mid-South U.S. Crop Production Systems, Katherine Suzanne French Dec 2025

Agricultural Practices’ Impact On Soil Health Indicators In Mid-South U.S. Crop Production Systems, Katherine Suzanne French

Graduate Theses and Dissertations

Soil health and regenerative agriculture are concepts gaining popularity across global agriculture systems. The effect of sustainable farming practices such as nutrient management, cover crops, adoption of no-tillage, and residue retention on crop yield, as well as environmental resilience, is increasingly being studied. This research aimed to evaluate the impacts of (i) cover cropping and nutrient management, and of (ii) soil sampling depth and timing on indicators of soil health in various mid-southern irrigated row crop systems. Soil health indicators like soil organic matter (SOM), carbon dioxide respiration (CDR), beta-glucosidase enzyme activity (BG), permanganate oxidizable carbon (POXC), and the soil …

Go to article

Beam Steering Control For A Small-Scale 5g Antenna Array, Omar Wagih Elkalesh Dec 2025

Beam Steering Control For A Small-Scale 5g Antenna Array, Omar Wagih Elkalesh

Graduate Theses and Dissertations

This thesis demonstrates the design, development, and validation of a full-scale adaptive beam steering control system of a small-scale 5G antenna array, designed and thoroughly evaluated with MATLAB Simulink. Overcoming the significant challenge of establishing strong wireless links in dynamic environments, the new system uses a new closed-loop control framework that can automatically detect the signal sources and automatically change the beam direction in real-time. The use of the adaptation is necessary to combat tracking errors that are caused by the mobility of the mobile users, environmental variations, as well as channel estimation errors, widespread in future wireless communication systems. …

Go to article

Predicting Stock Price Movement With Llm-Enhanced Tweet Emotion Analysis, An Vuong Dec 2025

Predicting Stock Price Movement With Llm-Enhanced Tweet Emotion Analysis, An Vuong

Graduate Theses and Dissertations

Accurately predicting short-term stock price movement remains a challenging task due to the market’s inherent volatility and sensitivity to investor sentiment. In this thesis, we present a published paper that discusses a deep learning framework integrating emo- tion features extracted from tweet data with historical stock price information to forecast significant price changes on the following day. We utilize Meta’s LLaMA 3.1-8B-Instruct model to preprocess tweet data, thereby enhancing the quality of emotion features derived from three emotion analysis approaches: a transformer-based DistilRoBERTa classifier from the Hugging Face library and two lexicon-based methods using National Research Council Canada (NRC) resources. …

Go to article

Optical Characterization Of Ge And Gesn Grown In Aspect Ratio Trapping Structures For Enhanced Emitter Performance, Abdulla Said Ali Dec 2025

Optical Characterization Of Ge And Gesn Grown In Aspect Ratio Trapping Structures For Enhanced Emitter Performance, Abdulla Said Ali

Graduate Theses and Dissertations

Silicon (Si) began dominating the electronics industry since 1960s, the dominance came due to its affordability and well developed fabrication processes. But when it comes to light emitting devices and photonics devices in general, the indirect bandgap of Si has caused many scientists to search for alternative semiconductor materials that can deliver better light emission, materials that can also improve wavelengths detection which is also very limited when it comes to Si, reduce losses and greater light amplification, all this while still being affordable. Despite these limitations, Si has been widely used in Complementary Metal Oxide Semiconductor (CMOS) manufacturing and …

Go to article

Automation And Upgrading Of Custom Cold Wall Uhv-Cvd Reactor For Group Iv Material Development, Alexander Golden Dec 2025

Automation And Upgrading Of Custom Cold Wall Uhv-Cvd Reactor For Group Iv Material Development, Alexander Golden

Graduate Theses and Dissertations

Group IV alloys have shown promising results for infrared optoelectronic applications where III-V materials are currently dominant. The III-V material system has vacancies that Group IV materials can fill, most importantly complementary metal oxide semiconductor (CMOS) process compatibility making quality devices much cheaper and widely available compared to the current state of the art. Silicon, being an indirect band gap Group IV material, has poor performance in optoelectronic applications but is the dominant material for most of the electronic industry. However, when silicon is alloyed or substituted with other group IV materials, these optoelectronic properties can be enhanced and, in …

Go to article

Evidence Of Backwater Morphodynamics In Stratigraphy On Earth And Mars: Implications For Ancient Mars, Cory Mitchell Hughes Dec 2025

Evidence Of Backwater Morphodynamics In Stratigraphy On Earth And Mars: Implications For Ancient Mars, Cory Mitchell Hughes

Graduate Theses and Dissertations

Geologic evidence indicates that liquid water once flowed in abundance across the martian surface. For this reason, in the search for evidence of extraterrestrial life, Mars represents a prime target for scientific inquiry. River systems encode climatic and environmental signals in their deposits on Earth and Mars, often in fan-shaped landforms (e.g., alluvial fans, river deltas, etc.). Distal fan-shaped deposits (i.e., river deltas), in particular, act as a relatively high-fidelity record of dynamic geologic processes. As such, NASA frequently sends rovers to locations on Mars’ surface with hypothesized river deltas (e.g., Perseverance and Jezero crater). Problematically, there are many different …

Go to article

Solar Powered Conservation: The Effects Of Solar Photovoltaic Facilities On Avian Community Composition, Michael Christopher Ferrara Dec 2025

Solar Powered Conservation: The Effects Of Solar Photovoltaic Facilities On Avian Community Composition, Michael Christopher Ferrara

Graduate Theses and Dissertations

The expansion of solar photovoltaic energy infrastructure is transforming landscapes worldwide. Increasingly, facilities are planted with native vegetation, yet how vegetation characteristics and landscape variables shape wildlife communities in solar facilities remains poorly understood. Because avian populations have been in decline and are sensitive to habitat changes, understanding how solar facilities influence avian communities is important. In Chapter 1, we used autonomous recording units, local vegetation measurements, and land use and land cover data to quantify how avian community occupancy is influenced by solar cover in solar facilities and comparable reference sites. We used a Bayesian multi-species occupancy model to …

Go to article

Towards Vision-Brain Understanding At Scales: From Classical To Quantum Machine Learning Approaches, Xuan-Bac Nguyen Dec 2025

Towards Vision-Brain Understanding At Scales: From Classical To Quantum Machine Learning Approaches, Xuan-Bac Nguyen

Graduate Theses and Dissertations

In recent years, large-scale learning approaches such as unsupervised and self-supervised learning have revolutionized artificial intelligence. These methods enable machines to learn high-level representations without explicit human supervision, achieving remarkable success across vision, language, and multimodal tasks. However, such advances come at a cost—they rely on massive datasets, billions of parameters, and extensive computational resources. Despite these achievements, artificial systems still fall short of the remarkable learning efficiency of the human brain, which can infer, adapt, and generalize from limited experiences. This gap motivates a deeper exploration of how biological intelligence acquires knowledge and how these principles can inspire the …

Go to article

Predictive Modeling Of Listeria Monocytogenes Growth In Ready-To-Eat Meat Products, Manjari Singh Dec 2025

Predictive Modeling Of Listeria Monocytogenes Growth In Ready-To-Eat Meat Products, Manjari Singh

Graduate Theses and Dissertations

Listeria monocytogenes remains a significant food safety concern in ready-to-eat (RTE) meat and poultry products due to its persistence under refrigerated and stressed conditions. Predictive models play a vital role in risk assessment and food safety management; broth models are highly conservation & results in overprocessing. This thesis focuses on developing and validating meat-specific predictive models for Listeria monocytogenes growth and lag for RTE beef, pork, and poultry products. Matrix-specific secondary models were built to predict the maximum specific growth rate (μmax) of Listeria monocytogenes under varying environmental conditions, including temperature, pH, water activity, salt, organic acids, and nitrites. Growth …

Go to article

Study And Simulation Of Model-Reference Adaptive Control Via The Heffron-Philips Model For Enhanced Power System Stability, Laila Ayman Dallol Dec 2025

Study And Simulation Of Model-Reference Adaptive Control Via The Heffron-Philips Model For Enhanced Power System Stability, Laila Ayman Dallol

Graduate Theses and Dissertations

Improving power transmission and quality has been an area of research interest over the past decades and was improved using renewable energy resources, mainly wind, and solar generation. Renewable energy penetration into power grids aims to replace some traditional synchronous generators, affecting the overall power system with reduced gas emissions that help with climate change and pollution issues. However, renewable generation in power systems ranging from microgrids to national grids is subjected to large disturbances that affect its transient stability, which is measured by its critical clearing time. As increased amounts of renewable generation connected to power systems have resulted …

Go to article

Updating The Surface Geological Maps In The Boston Mountains Region Of Arkansas Using Subsurface Well Log Correlation And Lidar Mapping, Olivia Smith Dec 2025

Updating The Surface Geological Maps In The Boston Mountains Region Of Arkansas Using Subsurface Well Log Correlation And Lidar Mapping, Olivia Smith

Graduate Theses and Dissertations

The Pennsylvanian Atoka Formation are the primary surface outcrops of the Boston Mountains in northwest Arkansas. Where fully preserved in the Arkoma Basin to the south, the Atoka Formation exceeds 18,000 feet of sandstone and shale. It is mapped as a single “undifferentiated” unit on the surface geological map of Arkansas. This project correlates ten subsurface subdivisions of the Atoka Formation plus the Desmoinesian to the surface outcrops of the southern Boston Mountains. This produced a higher resolution surface map beyond the current Atoka Undifferentiated scheme. Recognition of the Atoka Formation subdivisions in surface outcrops using standard lithostratigraphic and biostratigraphic …

Go to article

Unraveling Nucleation Mechanisms In Membrane Distillation: A Molecular Dynamics And Enhanced Sampling Approach, Nazanin Abbasi Dec 2025

Unraveling Nucleation Mechanisms In Membrane Distillation: A Molecular Dynamics And Enhanced Sampling Approach, Nazanin Abbasi

Graduate Theses and Dissertations

Membrane distillation offers a promising route for desalination but remains limited by salt scaling, where mineral crystals nucleate and grow in concentrated brines. To inform strategies that mitigate scaling, this thesis investigates homogeneous nucleation in supersaturated aqueous NaCl solutions using molecular dynamics and enhanced sampling techniques. Several collective variables (CVs) are tested to capture the early stages of ion aggregation and crystal formation, including ion–ion coordination number, dehydration, and bond orientational order parameters Q₄ and Q₆. Enhanced sampling techniques, such as metadynamics and umbrella sampling, are employed to overcome the high free energy barriers associated with the formation of critical …

Go to article

Novel Radical Chemistry Of Electron-Rich Double Bonds, Claire Allyson Beard Dec 2025

Novel Radical Chemistry Of Electron-Rich Double Bonds, Claire Allyson Beard

Graduate Theses and Dissertations

This work delves into the reactivity of nucleophilic N-heterocyclic carbenes (NHCs) and their ability to form electron-rich alkene intermediates, which can undergo various chemical transformations depending on the nature of the electrophile. Initial reports by Hahn disclosed the NHC promoted radical nature of electron-rich N,N’-benzyltetraazafulvalenes, which led to 2,2’- biimidazoles. We found that N-aryl substituted pyridinium salts also underwent rapid a similar NHC promoted dimerization when exposed to a base, producing 2,2’-bipyridyls. This is likely due to the adept radical nature of the bipyridinylidene intermediate for C-Naryl bond scission. We have also uncovered an unprecedented C-N bond scission of Breslow-like …

Go to article

Contextual Embedding Using Machine Learning For Cybersecurity: Access Control And Application, Thanh Bui Dec 2025

Contextual Embedding Using Machine Learning For Cybersecurity: Access Control And Application, Thanh Bui

Graduate Theses and Dissertations

Access control is a well-established challenge in cybersecurity, with significant research focused on enhancing system autonomy and accuracy across various scenarios. Access control rules can be designed based on users’ roles, attributes, or relationships requesting access to specific resources. However, despite their benefits, these models still require human oversight. As systems expand and grow, it becomes increasingly complex for administrators to maintain precise access control rules, often necessitating extensive system updates or even a complete overhaul. This dissertation introduces a novel approach that leverages contextual embedding for user information to enable the system to autonomously authorize user requests for resources. …

Go to article

Ring-Opening And Monofunctionalization Study On Spiro Cyclic Aminals And N, O-Acetals System Via Nitrogen-Centered Radical Intermediates, Chujun Zhou Dec 2025

Ring-Opening And Monofunctionalization Study On Spiro Cyclic Aminals And N, O-Acetals System Via Nitrogen-Centered Radical Intermediates, Chujun Zhou

Graduate Theses and Dissertations

The nitrogen-centered radical has been drawing increasing attention among researchers. Radical polarity is one of the key factors in studying nitrogen-centered radicals’ reactivity, and the functional groups attached to the nitrogen atom are of vital importance. Among different types of nitrogen-centered radicals, aminium radical and metal aminium radical are electrophilic due to the positive charge on the intermediate. The sulfonamide radical, amidyl radical, and cyanide amine radical are also electrophilic since the nitrogen is adjacent to electron electron-withdrawing group. The neutral amino radical, iminyl radical, masked nitrogen radical are nucleophilic because of the electron-donating groups. After classifying the nitrogen-centered radical, …

Go to article

Multiple Aspects Of Environmental Temperature Have Complex Interactions With Avian Parental Behavior To Influence Offspring Outcomes, William H. Kirkpatrick Dec 2025

Multiple Aspects Of Environmental Temperature Have Complex Interactions With Avian Parental Behavior To Influence Offspring Outcomes, William H. Kirkpatrick

Graduate Theses and Dissertations

Anthropogenic climate change is projected to shift not only average temperatures across the globe, but also the amount of thermal variation experienced in the natural environment. Exposure to shifts in averages and a wider range of thermal experiences will cause shifts in wildlife behavior. In avian species, behaviors exhibited during reproduction have implications for the well-being of their offspring. Here, I explore specific incubation behaviors in four species: Scaled Quail (Callipepla squamata), Northern Bobwhite Quail (Colinus virginianus), Tree Swallows (Tachycineta bicolor), and Eastern Bluebirds (Sialia sialis). I examined constancy, or the percentage of time on the nest, and off-bout duration, …

Go to article

When The Grid Goes Dark: A Digital Forensics Study Of Industrial Control System Cyberattacks, Katie Kettler Dec 2025

When The Grid Goes Dark: A Digital Forensics Study Of Industrial Control System Cyberattacks, Katie Kettler

Graduate Theses and Dissertations

Industrial Control Systems (ICS) and Operational Technology (OT) maintain the grid, ensure water safety, and keep transportation running. Because they influence nearly every aspect of daily life, these systems have become prime targets for cyberattacks. The need for this research arises from the fact that when ICS and OT systems are compromised, the consequences go beyond data loss, and they can directly disrupt communities and endanger public safety. This thesis introduces digital forensics fundamentals and explains how investigations in ICS environments differ from those in traditional IT environments. This work then examines major attacks, including Stuxnet, the Ukrainian Grid Attacks …

Go to article

Two Topics In Survival Analysis: Restricted Distance Covariance Test For Non-Proportional Hazard And A New Estimation For Dropout Rate, Ruizhe Yin Dec 2025

Two Topics In Survival Analysis: Restricted Distance Covariance Test For Non-Proportional Hazard And A New Estimation For Dropout Rate, Ruizhe Yin

Graduate Theses and Dissertations

When treatment effects change over time, standard statistical methods, such as the log-rank test and the Cox proportional hazards model, may give misleading results. This dissertation presents the restricted distance covariance (rdcov) test, a nonparametric method that compares survival curves between groups within a chosen study period [0, τ] using right censored data. The statistic measures the dependence between pre-specified group labels and survival times using pairwise distances from Kaplan-Meier estimates. Our method does not rely on the proportional hazards assumption, and it equals zero only when survival functions are identical across groups. Thus, this test can be applied to …

Go to article

Design Of Phase Shifter And True Time Delay Gan Mmics For High-Temperature Ku-Band Applications, Michael Lee Thompson Dec 2025

Design Of Phase Shifter And True Time Delay Gan Mmics For High-Temperature Ku-Band Applications, Michael Lee Thompson

Graduate Theses and Dissertations

This thesis presents the design, implementation, and high-temperature characterization of Gallium Nitride (GaN) monolithic microwave integrated circuits (MMICs) developed for beamforming and phased-array systems operating in the Ku-band (12-14 GHz). Two integrated circuits were designed: a 3-bit digital phase shifter and a 3-bit true time delay (TTD), both optimized for high linearity, low insertion loss, low phase error, and stable operation at elevated temperatures up to 300°C. Developing both a phase shifter and a TTD enabled a direct comparison of their beamforming performance and the evaluation of beam squint effects over frequency, a critical factor in wideband array design. The …

Go to article

A Gis Method To Analyze Trajectories Of Gross Changes Through Multiple Temporal Resolutions With An Application To Brazilian Soybean Cultivation, Thomas Mumuni Bilintoh, Robert Gilmore Pontius, Gustavo Oliveira, Julia Shimbo Dec 2025

A Gis Method To Analyze Trajectories Of Gross Changes Through Multiple Temporal Resolutions With An Application To Brazilian Soybean Cultivation, Thomas Mumuni Bilintoh, Robert Gilmore Pontius, Gustavo Oliveira, Julia Shimbo

Geography

Our manuscript presents Trajectory Analysis, which is a novel method to characterize patterns of gross changes during a time series of raster maps. This manuscript is our first article to elaborate on concepts and features of Trajectory Analysis that are necessary to analyze more than two time intervals. The method computes up to eight trajectories and three components of change: Quantity, Exchange, and Alternation. Quantity is the absolute net change between the two time points that bound the temporal extent. Exchange is the additional gross change between the same two time points. Alternation is the additional gross change that derives …

Go to article

Using Artificial Intelligence (Ai) To Improve The Identification Of Gifted Students, Hamid Ahmad Almalki Dec 2025

Using Artificial Intelligence (Ai) To Improve The Identification Of Gifted Students, Hamid Ahmad Almalki

Graduate Theses and Dissertations

The present study examined teachers' perceptions of using artificial intelligence (AI) to identify gifted students. The issue is clearly relevant today; several schools continue to grapple with unfair practices and irregular assessment processes, most notably for culturally and linguistically different students. Conventional keys and expert opinion remain essential, but their limitations have been acknowledged. Newer research suggests that AI could actually improve the system by making it more efficient and able to pool data that originally surfaced biases. At the same time, issues of transparency, data privacy, and the ethics of decisions driven by AI remain unresolved. To gain an …

Go to article

Evaluating Toggle-Out Testing For Pedicle Screw Failure Prediction, Rebecca Elayne Shelby Dec 2025

Evaluating Toggle-Out Testing For Pedicle Screw Failure Prediction, Rebecca Elayne Shelby

Master of Engineering Theses

Pedicle screw loosening is a common failure mode in spinal instrumentation, particularly in osteoporotic bone, where cancellous strength averages 0.944 MPa compared to 1.5-7.8 MPa in healthy bone. This study evaluated toggle-out testing for predicting pedicle screw failure and examined how material strength affects failure behavior. Polyurethane blocks underwent progressive cyclic loading from 100 N to 400 N in 100 N increments, with displacement recorded throughout. A 200 N load reflects forces typically carried by a screw during walking in an osteoporotic patient, placing the test range within expected clinical loading. Displacement behavior and stiffness decay were quantified using a …

Go to article

Landfast Legalities And Institutional Erosion: Assembling Alaska Coastal Governance At The Melting Edge Of The Anthropocene, Nicholas J. Parlato Dec 2025

Landfast Legalities And Institutional Erosion: Assembling Alaska Coastal Governance At The Melting Edge Of The Anthropocene, Nicholas J. Parlato

Interdisciplinary Studies

Anthropocene discourses raise questions about the political agency of nonhuman nature and the error of human exceptionalism, blurring distinctions between the subjects and objects of environmental governance. In the Arctic, the environment’s capacity to determine how and what kind of activities can take place has long existed in tension with the legal and political systems imposed by colonial powers. How can governance and decision-making for destabilized socionatural communities be reconceived from a more-than-human perspective? Through the lenses of assemblage thinking, posthumanism, and Indigenous law, this dissertation interrogates conventional approaches to understanding and governing the Arctic and, focusing on the history …

Go to article

Digital Communications Between Firms And Investors: Impact Of Explanatory Responses On Investor Engagement In Online Financial Q&A, Runyu Wang, Zili Zhang, Keng Siau, Ziqiong Zhang Dec 2025

Digital Communications Between Firms And Investors: Impact Of Explanatory Responses On Investor Engagement In Online Financial Q&A, Runyu Wang, Zili Zhang, Keng Siau, Ziqiong Zhang

Research Collection School Of Computing and Information Systems

The emerging trend of digital communications between firms and investors through online question-and-answer (Q&A) platforms is recognized as a vital strategy for managing investor relations, contributing to enhanced market efficiency and information transparency through increased information exchange. Potential investors can seek responses from firm managers to address their information needs, thereby mitigating market uncertainties. To provide foundational insights, we conduct a survey of investors to assess their awareness, usage, and perceptions of firm-investor Q&A platforms. In the subsequent empirical study, we specifically focus on the substance of managers’ responses, which are primarily aimed at clarifying firm events or information. In …

Go to article

Backdoorllm: A Comprehensive Benchmark For Backdoor Attacks And Defenses On Large Language Models, Yige Li, Hanxun Huang, Yunhan Zhao, Xingjun Ma, Jun Sun Dec 2025

Backdoorllm: A Comprehensive Benchmark For Backdoor Attacks And Defenses On Large Language Models, Yige Li, Hanxun Huang, Yunhan Zhao, Xingjun Ma, Jun Sun

Research Collection School Of Computing and Information Systems

Generative large language models (LLMs) have achieved state-of-the-art results on a wide range of tasks, yet they remain susceptible to backdoor attacks: carefully crafted triggers in the input can manipulate the model to produce adversaryspecified outputs. While prior research has predominantly focused on backdoor risks in vision and classification settings, the vulnerability of LLMs in open-ended text generation remains underexplored. To fill this gap, we introduce BackdoorLLM1 , the first comprehensive benchmark for systematically evaluating backdoor threats in text-generation LLMs. BackdoorLLM provides: (i) a unified repository of benchmarks with a standardized training and evaluation pipeline; (ii) a diverse suite of …

Go to article

Griffin: Effective Token Alignment For Faster Speculative Decoding, Shijing Hu, Jingyang Li, Xingyu Xie, Zhihui Lu, Kim-Chuan Toh, Pan Zhou Dec 2025

Griffin: Effective Token Alignment For Faster Speculative Decoding, Shijing Hu, Jingyang Li, Xingyu Xie, Zhihui Lu, Kim-Chuan Toh, Pan Zhou

Research Collection School Of Computing and Information Systems

Speculative decoding accelerates inference in large language models (LLMs) by generating multiple draft tokens simultaneously. However, existing methods often struggle with token misalignment between the training and decoding phases, limiting their performance. To address this, we propose GRIFFIN, a novel framework that incorporates a token-alignable training strategy and a token-alignable draft model to mitigate misalignment. The training strategy employs a loss masking mechanism to exclude highly misaligned tokens during training, preventing them from negatively impacting the draft model’s optimization. The token-alignable draft model introduces input tokens to correct inconsistencies in generated features. Experiments on LLaMA, Vicuna, Qwen and Mixtral models …

Go to article

Sopo: Text-To-Motion Generation Using Semi-Online Preference Optimization, Xiaofeng Tan, Hongsong Wang, Xin Geng, Pan Zhou Dec 2025

Sopo: Text-To-Motion Generation Using Semi-Online Preference Optimization, Xiaofeng Tan, Hongsong Wang, Xin Geng, Pan Zhou

Research Collection School Of Computing and Information Systems

Text-to-motion generation is essential for advancing the creative industry but often presents challenges in producing consistent, realistic motions. To address this, we focus on fine-tuning text-to-motion models to consistently favor highquality, human-preferred motions—a critical yet largely unexplored problem. In this work, we theoretically investigate the DPO under both online and offline settings, and reveal their respective limitation: overfitting in offline DPO, and biased sampling in online DPO. Building on our theoretical insights, we introduce Semi-online Preference Optimization (SoPo), a DPO-based method for training text-to-motion models using “semi-online” data pair, consisting of unpreferred motion from online distribution and preferred motion in …

Go to article

First Prev Next Last