32584 research outputs found
Sort by
Fair Influence Maximization with Reverse Influence Sampling Boosted Multi-Objective Genetic Algorithms
No full textInfluence Maximization (IM) focuses on identifying a small set of influential nodes (seed nodes) in social networks to maximize the spread of information, ideas, or behaviors efficiently. Traditional IM methods prioritize overall influence spread, neglecting fairness considerations regarding who benefits from that spread. This oversight might result in particular disadvantaged demographic groups being omitted from seed selection and thus receiving critical information later, or not at all. This raises ethical concerns in scenarios involving health, education, or employment resources. Fair Influence Maximization (FIM) seeks to ensure equitable diffusion across predefined groups. While early FIM approaches, such as max–min fairness, address disparities in a simple way, they fail to capture subtler inequities. Recent frameworks incorporate statistical measures such as the Gini coefficient to evaluate inequality across influenced nodes. This work introduces FIMMOGA++, an enhanced genetic algorithm that integrates multi-objective optimization and fairness-aware design. It combines traditional IM heuristics with Reverse Influence Sampling (RIS) and the Non-dominated Sorting Genetic Algorithm III (NSGA-III) to simultaneously optimize influence spread and multiple fairness criteria. By producing a Pareto-optimal set of solutions, FIMMOGA++ provides balanced strategies that advance both theoretical understanding and practical implementation of fair influence maximization in large-scale networks
Enhancing LLM’s Mathematical and Theorem Proving Abilities
No full textLarge language models (LLMs) are good at natural language tasks but are not up to the mark when it comes to mathematics and theorem-proving, as they rely on language patterns instead of understanding the problem and thinking through the solution. This project addresses issues such as a lack of exposure to structured datasets, difficulty with generating outputs that require multi-step reasoning, and limitations of short context. We fine-tune pre-trained LLMs on structured datasets like MATH, GSM8k, open-r1, deepseek-prover, and OpenBootstrappedTheorem. We integrate two software tools, Mathics and LEAN, and enhance reasoning through Chain-of-Thought (CoT). Additionally, we conduct the experiments using state of the art Mixture of Experts (MoE) and parameter efficient fine-tuning (PEFT) techniques such as LoRA and DoRA. The outcomes of this project are better model performance on complex math problems, and particularly on formal theorem-proving datasets, which is a comparatively understudied domain in recent LLM research. This takes a step toward developing and fine-tuning models that can handle challenging mathematical and logical domains
Empowering customer service with generative AI: enhancing agent performance while navigating challenges
Get PDFIntroduction. As large language models (LLMs), such as GPTs, become more intelligent, a key area of exploration is how these technologies can improve the customer experience. Contrary to common belief, many consumers, including Gen Z, prefer human-provided customer service, illustrating the importance of human-AI collaboration in the space. Method. By leveraging the author’s real-world knowledge of enterprise knowledge management and customer service delivery, we reviewed numerous literature about AI, knowledge management, and service design and synthesised practical insights for industry professionals to build a successful AI strategy. Analysis. We examined the gap between academic research on generative AI and how such solutions can be deployed in real-world enterprise contact centers. Results. We outline how companies are adopting generative AI, offer strategies for training AI models to assist with customer issues, discuss AI-powered localization, and share tips for facilitating human-AI collaboration. Conclusion. This research found that companies will require humans for effective customer service delivery because there is no complete substitute for human judgment in mission-critical settings. Despite this, companies can still see significant performance and customer satisfaction improvements by focusing on human-AI collaboration, and empowering human agents to focus on specialized tasks while the AI addresses rudimentary ones
The relationship between food insecurity and emotional eating during the coronavirus pandemic in northern California
Get PDFObjectives: This cross-sectional study examined the association between food insecurity and emotional eating and explored correlations between emotional eating and dietary behavior in the U.S during the pandemic. Methods: Participants (N = 515, 77.1 % female, 44.1 ± 14.7 years old) completed a validated online survey between August and November of 2020 including questions about food security and eating behaviors. A multivariable linear regression model was used to evaluate the association between food insecurity and emotional eating. Pearson correlations were calculated to assess the relationship between emotional eating and dietary behavior Results: Overall, 31.3 % of the participants experienced food insecurity. Food insecurity was associated with higher emotional eating in the crude model (beta = 0.33; 95 % confidence level [CL]: 0.15, 0.52), but not in the fully adjusted regression model (beta = 0.15; 95 % CL: −0.06, 0.37). Intakes of sugary snacks (r = 0.16, p \u3c 0.01) and salty snacks (r = 0.14, p \u3c 0.01) were weakly positively correlated with higher emotional eating scores, while the intakes of vegetables (r = −0.13, p \u3c 0.01) and alcohol (r = −0.09, p = 0.03) were negatively correlated with emotional eating Conclusions: In this study, emotional eating was not associated with food insecurity and was weakly positively correlated with the intake of energy-dense foods. The relationship between food insecurity and emotional eating is complex and not yet clearly defined. Examination of longitudinal associations between emotional eating, food insecurity, and energy-dense foods is warranted
Spartan Daily, April 15, 2025
Get PDFVolume 164, Issue 32https://scholarworks.sjsu.edu/spartan_daily_2025/1031/thumbnail.jp
Chains of Equity: How Nationwide Injunctions Are Unconstitutional
Get PDFSince the early judicial beginnings in the English Courts of Chancery, the principles of equity have guided judicial remedies for centuries. Modern American courts inherited these principles through reception statutes passed by newly independent states, which adopted English common law, including the concept of equity as a means of providing narrowly tailored remedies to resolve specific injuries. This paper examines the historical foundations of equity and argues that nationwide injunctions, which extend beyond the immediate litigants to affect the general public, are inconsistent with these principles and, therefore, unconstitutional. A historical and doctrinal analysis of the English Court of Chancery, equity principles, and federal judiciary practices reveals that equitable relief has traditionally been limited to the parties involved rather than imposing broad remedies on the general public. The findings confirm that equitable relief is only legitimate when it conforms to the principles of equity, reinforcing the argument that nationwide injunctions exceed the constitutional limits of judicial authority
The Effects of Muscle Fatigue on Lower Extremities Biomechanics During Layup and Landing in Male Recreational Basketball Players
No full textMuscle fatigue is a known contributor to musculoskeletal injuries, yet few studies have examined. its effects during the lay–up phase in sports. This study evaluated how fatigue alters ankle, knee, and hip biomechanics during lay–up and landing. We hypothesized that fatigue would increase peak vertical ground reaction force (GRF), knee extension angle, and joint moments. Fourteen healthy males performed 3–step lay–ups and drop–jumps on force plates, before and after a fatigue protocol. Kinematic data were collected using an 8–camera Vicon system (100 Hz), and GRF data with two AMTI force plates (1000 Hz). Thirty–six reflective markers were placed on lower limb anatomical landmarks, and data were processed in Visual 3D. Paired t–tests (α = .05) in SPSS (V26.0) compared pre– and post–fatigue outcomes. For the joint kinematic outcomes, a significant difference was found in peak knee extension angle (degree) between pre-fatigue (6.75 ± 6.01) and post- fatigue (9.28 ± 6.52) conditions during the layup phase (p = 0.028). For the kinetic outcomes, a significant difference was found in peak ground reaction force (BW) between the pre-fatigue (3.31 ± 0.81) and post-fatigue (3.95 ± 1.05) conditions during the landing phase (p = 0.036). In addition, a significant difference was found in the peak negative joint work between the pre-fatigue (-0.34 ± 0.18) and post-fatigue ( -0.66 ± 0.43) conditions during the landing phase (p = 0.025). These results suggest fatigue impairs neuromuscular control and increases injury ris
Interactions Between a High Intensity Wildfire and an Atmopsheric Hydraulic Jump in the Case of the 2023 Lahaina Fire
Get PDFOn 8 August 2023, a grass fire that started in the city of Lahaina, Hawai’i, grew into the deadliest wildfire in the United States since 1918. This wildfire offers a unique opportunity to explore the impact of high heat output on an atmospheric hydraulic jump and a downslope wind event. We conducted two WRF-SFIRE simulations to investigate these effects: one incorporating fire–atmosphere feedback and the other without it. We compare these WRF-SFIRE simulations to all available data regarding the fire event, including high resolution satellite data, weather station data from other parts of Maui and the Hawaiian Archipelago, and official reports from the Maui Police Department, Fire Department, and the Fire Safety Research Institute. Our findings revealed that, in the uncoupled simulation, the hydraulic jump moved inland significantly earlier than in the coupled simulation. This altered the wind pattern near the fire front in the uncoupled simulation, accelerating its lateral spread. The results suggest that fire–atmosphere interactions and their influence on near-fire circulation may be more intricate than previously understood. Specifically, while fire-induced wind acceleration is often linked to enhanced fire spread, this study highlights that, in cases where the lateral fire spread is dominant, fire-induced circulation may reduce cross-flank flow and inhibit the fire growth
Greenwashing as Disinformation: News Media Sentiments to Deceptive Environmental Marketing Claims
No full textGreenwashing, the practice of misleading consumers about a company’s environmental practices, has become a growing phenomenon as businesses seek to appeal to eco-conscious consumers. This thesis examines greenwashing as a form of disinformation through a content analysis of news media coverage surrounding three major greenwashing scandals: Volkswagen, Coca-Cola, and H&M. Using BrandWatch software, 300 news articles (100 per company) published within two years after each greenwashing lawsuit were collected. Sentiment analysis and variables grounded in Blame Attribution Theory were examined, alongside other variables: Use of Strong Accusatory Language Implying Disinformation and Impact score. The study found no significant differences in negative sentiment across cases, suggesting consistent media framing. However, the presence of Strong Accusatory Language Implying Disinformation was found to be correlated with negative sentiment. This research contributes to discussions on consumer protection, and corporate transparency, offering insights for consumers and marketers navigating the growing intersection of environmental responsibility and brand messaging. Media literacy education is encouraged as a means for consumers to combat greenwashing
The Influence of Maternal Size/Age Effects On The Physiological Responses of Adult Female Gopher Rockfish (Sebastes carnatus) To Ocean Acidification and Hypoxia
Get PDFClimate change is rapidly reshaping the chemistry of the ocean. Fishes living in California coastal waters are experiencing increased ocean acidification and hypoxia (OAH) due to more frequent and intense upwelling. Nearshore rockfish may be particularly threatened by these conditions due to their long generational times. However, it is unknown how OAH may impact maternal physiology and reproduction in these viviparous fish. To understand the physiological effects of OAH during gestation, adult female gopher rockfish, Sebastes carnatus, were exposed to a variety of combined OAH stress treatments during different gestational stages. Routine metabolic rate (RMR), maximum metabolic rate (MMR), blood hematocrit (Hct), hemoglobin (tHb), pCO2, HCO3-, Na+, K+, Cl-, and metabolites, were measured to assess physiological responses to OAH stress. Ovarian oxygen was measured to examine the ability to buffer embryos against low oxygen. Fish exposed to higher OAH stress displayed elevated blood Hct, tHb, pCO2 and HCO3-, and decreased MMR, indicating attempted compensation for low pH and hypoxia (with varying levels of success), at increased physiological costs. Fish showed signs of buffering their ovaries against hypoxia. Lastly, pregnancy altered Hct and RMR under OAH exposure and size/age did not have a consistent effect on maternal physiology. By evaluating responses of maternal physiology to OAH stress, we can better understand how climate change affects fecundity, larval condition, and survival, influencing nearshore fisheries in an ever-changing climate