University of Wisconsin–Stevens Point,University of Wisconsin System

Despite the global deployment of text-to-image (T2I) models, their safety frameworks are largely calibrated to a Western-centric default, creating significant vulnerabilities for the rest of the world. To embrace cultural pluralism and bring historically under-represented perspectives in T2I safety, we conduct localised community-centered red teaming studies in the Global South. Our two-fold approach prioritizes localization and participation, by focusing on secondary urban centers in these regions, and conducting community engagement and training workshops to contextualize local norms. As a result, we present PLACES, a dataset comprising over 26,000 examples of T2I model failures collected in partnership with universities in Ghana, Nigeria, and two regions of India (Karnataka and Punjab). Analysis of prompts collected reveals a wide-ranging diversity in socio-cultural and linguistic attributes, when compared to existing geography-agnostic crowdsourced red-teaming data. We observe unique adversarial patterns enabled by local cultural and linguistic nuances, and distinct clusters within region around specific themes, such as religion in India. Moreover, we uncover structural contextual gaps in existing safety frameworks by identifying novel harms showing normative dissonance (e.g., violating religious norms, ignoring local customs, and ominous symbolism). This work argues that expanding T2I safety requires moving beyond mere scale to incorporate deeply localised, participatory methodologies for data collection and contextualization. Content warning: This paper includes examples containing potentially harmful or offensive content.

Este documento reporta las lecciones aprendidas después de huracanes recopiladas en investigaciones llevadas a cabo por científicos de University of Florida/Institute of Food and Agricultural Sciences (UF/IFAS). También se incluyen observaciones de campo de profesionales como expertos forestales urbanos, científicos y arboricultores.

Level 1 or limited visual assessments are commonly conducted by utility and municipal arborists as a means of efficiently inspecting large populations of trees (Smiley et al., 2017). However, their effectiveness in identifying the trees most likely to fail in severe weather has not been documented. In this study, limited visual tree risk assessments (n = 2253) were conducted prior to a derecho wind event in Sheboygan, Wisconsin, and compared to post-storm response work orders to gauge their accuracy. Of the trees assessed, 8% were damaged during the storm (n = 169), including 3% which experienced whole tree failure (n = 67). Trees which appeared to have an elevated risk were further assessed with a Level 2 basic visual assessment (n = 38). Of these, 26% (n = 10) were damaged, including 11% (n = 4) which failed. Within the total street tree population of 17,846 trees, 2% of trees failed completely (n = 345) and 9% were damaged during the storm (n = 1603). To gauge the consequences of this damage to people and property, we surveyed residents whose trees were damaged or failed during the storm (n = 51), 43% of whom (n = 22) reported damage to property due to tree failure. Descriptions of property damage were limited to static targets, such as houses or parked cars. No personal injuries were reported in survey responses nor to local emergency response personnel the night of the storm. Our findings add to a growing body of literature demonstrating that tree failure is difficult to accurately predict.

Importance Fibromyalgia is characterized by chronic widespread pain that is often exacerbated by movement that interferes with daily activities. Development of effective treatments for movement-evoked pain is essential for improving function for individuals with fibromyalgia. Objective To evaluate whether the addition of transcutaneous electrical nerve stimulation (TENS) to outpatient physical therapy improves fibromyalgia-associated movement-evoked pain. Design, Setting, and Participants The Fibromyalgia TENS in Physical Therapy (FM-TIPS) study was a cluster-randomized clinical trial of participants with fibromyalgia at 28 outpatient PT clinics from 6 health care systems. Between February 1, 2021, and September 31, 2024, 958 participants were screened, 459 participants enrolled, and 384 completed baseline data collection, with final data collected in March 2025. Intervention Clinics were randomized to PT plus TENS (PT-TENS) and PT-only groups. Data were captured on days 1, 30, 60 (primary end point, randomized phase), 90, and 180. Participants in the PT-only group received TENS after day 60 (extension phase). TENS was applied to the upper and lower back with instructions to use 2 hours daily with parameters of modulating frequency of 2 to 125 Hz for 100 to 180 microseconds at a strong but comfortable intensity. Main Outcomes and Measures The primary outcome was a change in movement-evoked pain (scale of 0-10, with 0 indicating no pain and 10 indicating worst pain imaginable) from baseline to day 60 rated during a 5-times sit-and-stand task using a linear mixed-effects model. In addition, patient-reported improvement based on the Patient Global Impression of Change score and patient-reported adverse events were assessed. Results A total of 384 FM-TIPS participants (mean [SD] age, 53 [15] years; 351 [91%] female) completed baseline data collection (modified intention-to-treat), with 191 individuals in PT-TENS group and 193 in PT-only group. Movement-evoked pain at day 60 during TENS treatment was significantly lower in the PT-TENS group compared with the PT-only group (group mean difference, −1.2; 95 CI, −1.6 to −0.7; d = 0.46). A dose-response effect for TENS was observed, with more participants in the PT-TENS group reporting improvement on the Patient Global Impression of Change (120 [72%] vs 86 [51%], P = .001) and a 30% or greater reduction in movement-evoked pain in responder analysis (66 of 161 [41%] vs 22 of 169 [13%]; P < .001). At day 180, 217 respondents (81%) found TENS helpful and 147 (55%) used TENS daily. There were no serious adverse events, and 109 of 358 (30%) experienced minor adverse events during the entire 6 months of the study. Conclusions and Relevance In this cluster randomized clinical trial of TENS in fibromyalgia, TENS meaningfully reduced movement-evoked pain and remained effective for 6 months. This study’s results suggest that TENS is a safe, inexpensive, and readily available treatment for fibromyalgia. Trial Registration ClinicalTrials.gov Identifier: NCT04683042

Disagreement in annotation is a common phenomenon in the development of NLP datasets and serves as a valuable source of insight. While majority voting remains the dominant strategy for aggregating labels, recent work has explored modeling individual annotators to preserve their perspectives. However, modeling each annotator is resource-intensive and remains underexplored across various NLP tasks. We propose an agreement-based clustering technique to model the disagreement between the annotators. We conduct comprehensive experiments in 40 datasets in 18 typologically diverse languages, covering three subjective NLP tasks: sentiment analysis, emotion classification, and hate speech detection. We evaluate four aggregation approaches: majority vote, ensemble, multi-label, and multitask. The results demonstrate that agreement-based clustering can leverage the full spectrum of annotator perspectives and significantly enhance classification performance in subjective NLP tasks compared to majority voting and individual annotator modeling. Regarding the aggregation approach, the multi-label and multitask approaches are better for modeling clustered annotators than an ensemble and model majority vote. The dataset is publicly available in GitHub: https://github.com/Tadesse-Destaw/Beyond-Majority-Voting.