- Topic Modeling
- Speech and dialogue systems
- Sexual Assault and Victimization Studies
- Acute Ischemic Stroke Management
- Cerebrovascular and Carotid Artery Diseases
- Natural Language Processing Techniques
- Recommender Systems and Techniques
- Multi-Agent Systems and Negotiation
- Advancements in Semiconductor Devices and Circuit Design
- Semiconductor materials and devices
- Law in Society and Culture
- Peripheral Artery Disease Management
- Jury Decision Making Processes
- Dermatology and Skin Diseases
- Data Quality and Management
- Data Mining Algorithms and Applications
- Calibration and Measurement Techniques
- Perovskite Materials and Applications
- Bacteriophages and microbial interactions
- Atmospheric Ozone and Climate
- Alcohol Consumption and Health Effects
- Advanced Graph Neural Networks
- Liver Disease Diagnosis and Treatment
- Conducting polymers and applications
- Cannabis and Cannabinoid Research
University of Mississippi
2024
University of California, Berkeley
2024
TESLA (Czechia)
2024
Manchester Metropolitan University
2024
Google (United States)
2022-2023
University of Chicago
2023
University of Chicago Medical Center
2023
The Royal Melbourne Hospital
2022
The University of Queensland
2020-2022
Swansea University
2017-2021
Importance Recent large infarct thrombectomy trials used heterogeneous imaging modalities and time windows for patient selection. Noncontrast computed tomographic (CT) scan is the most common stroke approach. It remains uncertain whether effective patients with infarcts identified using noncontrast CT alone within 24 hours of onset. Objective To evaluate effect in a on Design, Setting, Participants Open-label, blinded–end point, bayesian-adaptive randomized trial interim analyses early...
Reinforcement learning from human feedback (RLHF) has proven effective in aligning large language models (LLMs) with preferences. However, gathering high-quality preference labels can be a time-consuming and expensive endeavor. RL AI Feedback (RLAIF), introduced by Bai et al., offers promising alternative that leverages powerful off-the-shelf LLM to generate preferences lieu of annotators. Across the tasks summarization, helpful dialogue generation, harmless RLAIF achieves comparable or...
Task-oriented dialogue (TOD) systems are required to identify key information from conversations for the completion of given tasks. Such is conventionally specified in terms intents and slots contained task-specific ontology or schemata. Since these schemata designed by system developers, naming convention not uniform across tasks, may convey their semantics effectively. This can lead models memorizing arbitrary patterns data, resulting suboptimal performance generalization. In this paper,...
LLM-powered personalization agent systems employ Large Language Models (LLMs) to predict users’ behavior from their past activities. However, effectiveness often hinges on the ability effectively leverage extensive, long user historical data due its inherent noise and length of such data. Existing pre-trained LLMs may generate summaries that are concise but lack necessary context for downstream tasks, hindering utility in systems. To address these challenges, we introduce Reinforcement...
Significance Substitution of cysteine by selenocysteine is held responsible for the increased performance many enzymes: The higher activity [NiFeSe]-hydrogenases compared with their [NiFe] counterparts often attributed to Sec replacement one active-site ligand. Replacing each four residues in an O 2 -tolerant [NiFe]-hydrogenase shows that this substitution alone does not overcome inability evolve H a characteristic group 1d hydrogenases. A nonbridging lying on direct path between Ni and...
Zero/few-shot transfer to unseen services is a critical challenge in task-oriented dialogue research. The Schema-Guided Dialogue (SGD) dataset introduced paradigm for enabling models support any service zero-shot through schemas, which describe APIs natural language. We explore the robustness of systems linguistic variations schemas by designing SGD-X - benchmark extending SGD with semantically similar yet stylistically diverse variants every schema. observe that two top state tracking fail...
Raghav Gupta, Harrison Lee, Jeffrey Zhao, Yuan Cao, Abhinav Rastogi, Yonghui Wu. Proceedings of the 2022 Conference North American Chapter Association for Computational Linguistics: Human Language Technologies. 2022.
Literature is sparse regarding men’s attitudes towards male sexual assault and the role that sexuality of those involved may have. Despite high prevalence chemsex GHB (gamma-hydroxybutyrate) participation among men who have sex with men, no study has yet investigated such. Utilising a community sample 141 UK participants were randomly assigned into one six conditions based on victim (heterosexual or homosexual) drug used present during (chemsex, chemical submission, drugs). All completed...
Abstract Alcohol intoxication causes serious diseases, whereas current treatments are mostly supportive and unable to remove alcohol efficiently. Upon consumption, is sequentially oxidized acetaldehyde acetate by the endogenous dehydrogenase aldehyde dehydrogenase, respectively. Inspired metabolism of alcohol, a hepatocyte‐mimicking antidote for through codelivery nanocapsules oxidase (AOx), catalase (CAT), (ALDH) liver, where AOx CAT catalyze oxidation acetaldehyde, while ALDH catalyzes...
MALT1 regulates immunity and inflammation in multiple ways, while its role rheumatoid arthritis (RA) is obscure. This study aimed to investigate the relationship of with disease features, treatment outcome, as well effect on Th1/2/17 cell differentiation underlying molecule mechanism RA.Totally 147 RA patients were enrolled. Then their blood Th1, Th2, Th17 cells detected by flow cytometry. Besides, PBMC expression was before (baseline), at week (W) 6, W12, W24. 30 osteoarthritis health...
Background Certain anti‐ HIV drugs alone or in combination are often associated with liver damages, which frequently worsened by alcohol consumption. We previously found an endoplasmic reticulum ( ER ) stress mechanism for the drug‐ and alcohol‐induced hepatic injuries animal models vitro hepatocytes. However, it is unknown whether induce similar cellular responses nonparenchymal cells. Methods Primary mouse hepatocytes PMH ), kupffer cells KC hepatocellular stellate HSC were freshly...
Endoplasmic reticulum (ER) stress and unfolded protein response (UPR) are involved in anti‐human immunodeficiency virus (HIV) drugs alcohol‐induced liver disease a significant number of patients infected with HIV. However, the precise mechanism by which alcohol cause ER remains elusive. We found that ritonavir‐boosted lopinavir (RL) activated two canonical UPR branches without activation third activating transcription factor 6 (ATF6) branch either HepG2 cells or primary mouse hepatocytes. In...
Victims of intimate partner violence (IPV) face significant barriers to having their complaints believed both when initially reporting experiences and giving evidence at trial. This is especially the case they have been sexually assaulted by partner. These stem not only from misperceptions about what IPV is, but also due a mismatch between features sexual assault in stereotypic expectations 'real' rape is—a violent surprise attack stranger an outside location. We examine research on schemas...
Jeffrey Zhao, Yuan Cao, Raghav Gupta, Harrison Lee, Abhinav Rastogi, Mingqiu Wang, Hagen Soltau, Izhak Shafran, Yonghui Wu. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.
Background We previously developed enzyme nanoparticles ( ENP ) of alcohol metabolism. This study was to evaluate protective effects facilitated removal blood and/or acetaldehyde on anti‐ HIV drugs and alcohol‐induced liver injuries. Methods were prepared for degrading completely 1) or partially into 2), which applied mice acute binge chronic‐binge feeding in the presence antivirals (ritonavir lopinavir). Liver pathologies examined assess . Results In model, 1 2 reduced concentration (BAC)...
Food insecurity is a growing problem on college campuses. By following user-centered design process, our team envisioned how can be tool for enacting social justice. posing the question "How might we support students challenged by food insecurity?" designed product called Locker that paired an online ordering system with collection of lockers to discreetly and conveniently retrieve items from campus pantry. In this experience report, introduce campuses, discuss issue connects concerns...
Jurors are less likely to believe a victim of rape when the events she describes inconsistent with stereotype and appear more consistent what occurs in consensual sex. This research investigated whether presenting stereotype-consistent early victim's testimony can lead jurors evaluate other described as depicting rape. In Study 1, convenience sample recruited at university (N = 217) watched video which assault was presented first or last. Participants also evaluated degree depicted sex they...
Introduction Acute large vessel occlusion (LVO) can be secondary to thromboembolism or intracranial atherosclerotic disease (ICAD). Data on LVO management due underlying ICAD are scarce. We hypothesized that patients with have worse clinical outcomes following mechanical thrombectomy (MT) than those without ICAD. Methods performed a retrospective analysis of consecutive who underwent MT for in academic comprehensive stroke center between January 14, 2018 and October 24, 2021. Presence at...
Conversational recommendation systems (CRS) aim to recommend suitable items users through natural language conversation. However, most CRS approaches do not effectively utilize the signal provided by these conversations. They rely heavily on explicit external knowledge e.g., graphs augment models’ understanding of and attributes, which is quite hard scale. To alleviate this, we propose an alternative information retrieval (IR)-styled approach item task, where represent conversations as...
The rise of time-domain astronomy including electromagnetic counterparts to gravitational waves, microlensing, explosive phenomena, and even astrometry with Gaia, are showing the power need for surveys high-cadence, large area, long time baselines study transient universe. A constellation SmallSats or CubeSats providing wide, instantaneous sky coverage down 21 Vega mag at optical wavelengths would be ideal addressing this need. We assembling CuRIOS-ED (CubeSats Rapid Infrared Optical...
LLM-powered personalization agent systems employ Large Language Models (LLMs) to predict users' behavior from their past activities. However, effectiveness often hinges on the ability effectively leverage extensive, long user historical data due its inherent noise and length of such data. Existing pretrained LLMs may generate summaries that are concise but lack necessary context for downstream tasks, hindering utility in systems. To address these challenges, we introduce Reinforcement...