Riverside City College

ABSTRACT Objectives In this paper, we examine heterogeneity in popular perceptions of inflation in the United Kingdom and the United States. We examine heterogeneity both in terms of perceptions of the rate of inflation and, also, in perceptions of the causes of inflation. Methods We analyze opinion data from the United Kingdom (2023) and the United States (2024), which examines perceptions of the rate of inflation and its causes. Results In contrast to what might be expected given the rise in affective partisanship, we find that individual circumstances have a larger effect on driving perceptions of the rate of inflation than partisanship in both countries. Perceptions of inflation rate vary strongly and consistently by one's income: less well‐off voters in both countries see inflation as being much higher than do the better‐off. Over and above this pattern in perceptions, we also see that different voters hold different understandings of what is causing inflation. Heterogeneity in how voters understand how the economy works is distinct from that found in the literature to date. Conclusions Voters differ a great deal over what they believe to be the causes of inflation and see strong effects of ideological leanings. One implication of this finding of heterogeneity is that “one size fits all” communication strategies by parties or government, which focus on aggregate levels of inflation, are only likely to appeal to a subset of the population.

Eco-driving strategy for connected and automated electric vehicles (CAEVs) shows great potential to improve traffic efficiency and reduce energy consumption. However, a fixed time-to- energy weight coefficient was usually assumed in existing studies, which often neglect considering different driver preferences (i.e., energy or time optimal). To address this problem, this paper proposes a deep learning based eco-driving strategy to predict the optimal acceleration while accounting for driver preferences. First, the eco-driving for CAEVs is formulated as a constrained optimization problem, and the optimal trajectory dataset is generated by the Latin hypercube sampling (LHS) method with nonlinear programming (NLP) solvers. Second, a novel imitation learning (IL) framework with a scenario encoder is proposed for mapping the optimal trajectory, and physical knowledge is fused to constrain the infeasible action of IL in rare cases. Numerical simulations demonstrate that the proposed method provides an accurate approximation of the optimal solution with different driver preferences. Furthermore, interpretability analysis is employed to reveal the causal relationships between network inputs and outputs. The code is provided by the Github repository: https://github.com/cwan-ucr/MLTPA-DP.git.

To develop a self-supervised learning (SSL) model that classifies optic disc phenotypes in primary open angle glaucoma (POAG) and explores novel phenotypic patterns with optic disc photographs (ODPs). We collected 850 ODPs from patients with POAG and applied data augmentation to address class imbalances, yielding 10,493 images. Using the DINO Vision Transformer as the backbone model, we trained an SSL model to extract 2048-dimensional latent features. These features were used for both supervised classification of six known phenotypes and unsupervised clustering. Classification performance was evaluated with Random Forest and XGBoost models. UMAP (Uniform Manifold Approximation and Projection) was used for dimensionality reduction and feature visualization, and attention maps were generated for model interpretability. The DINO-EYE model features enabled phenotype classification with 91% accuracy with Random Forest and 92.1% after merging clinically similar phenotypes. Unsupervised clustering revealed coherent groupings, particularly for concentric thinning and extensive Peripapillary Atrophy (PPA), though no new phenotypes were unanimously confirmed by clinicians. The proposed model outperformed the RETFound SSL model in phenotype classification and demonstrated interpretable attention regions consistent with expert criteria. Our DINO-EYE effectively extracts clinically meaningful features from fundus images and enables accurate classification of optic disc phenotypes in POAG. It surpasses existing SSL models in performance and interpretability, offering promise for real-world glaucoma decision support and individualized care planning.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33140-1.

Horizon scanning is intended to identify opportunities and threats associated with technology, regulatory, and social change. Here, we report the results of a new horizon scan based on inputs of an international group of 33 participants, focusing on future issues arising from the military use of artificial intelligence (AI) for augmenting human performance. The final list of 12 issues includes topics spanning from the political (educating and training individuals to accept and work with AI), to the regulatory (issues of consent to human-AI teaming and hybridization), to security (the hackability of neural devices that connect to AI), to philosophical (the nature and phenomenology of brain-to-brain interfaces). The early identification of such issues is relevant to researchers, policymakers, military practitioners, and the wider public.

Real-world Vehicle-to-Everything (V2X) cooperative perception systems often operate under heterogeneous sensor configurations due to cost constraints and deployment variability across vehicles and infrastructure. This heterogeneity poses significant challenges for feature fusion and perception reliability. To address these issues, we propose HeCoFuse, a unified framework designed to enable effective cooperative perception across diverse sensor combinations, a unified framework designed for cooperative perception across mixed sensor setups-where nodes may carry Cameras (C), LiDARs (L), or both. By introducing a hierarchical fusion mechanism that adaptively weights features through a combination of channel-wise and spatial attention, HeCoFuse can tackle critical challenges such as cross-modality feature misalignment and imbalanced representation quality. In addition, an adaptive spatial resolution adjustment module is employed to balance computational cost and fusion effectiveness. To enhance robustness across different configurations, we further implement a cooperative learning strategy that dynamically adjusts fusion type based on available modalities. Experiments on the real-world TUMTraf-V2X dataset demonstrate that HeCoFuse achieves 43.22% 3D mAP under the full sensor configuration (LC+LC), outperforming the CoopDet3D baseline by 1.17%, and reaches an even higher 43.38% 3D mAP in the L+LC scenario, while maintaining 3D mAP in the range of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$21.74 \%-43.38 {\%}$</tex> across nine heterogeneous sensor configurations. These results, validated by our first-place finish in the CVPR 2025 DriveX challenge, establish HeCoFuse as the current state-of-the-art on TUM-Traf V2X dataset while demonstrating robust performance across diverse sensor deployments. Code is available at https://github.com/ChuhengWei/HeCoFuse.

This study elucidates PM2.5 exposure disparities in a socioeconomically diverse air basin that is heavily burdened by air pollution. A novel spatial clustering approach is applied to classify the microenvironments of more than 900,000 high temporal resolution personal exposure data points. Results from the study indicate that participants from the lowest socioeconomic status community experienced overall higher personal exposures over consecutive 24-hr monitoring periods, despite high participant mobility and low variability in ambient PM2.5 during the study. Our inclusive monitoring protocol minimizes participant fatigue and is well-suited for real-time, long-term characterization of PM2.5 exposure disparities in underserved communities.

Community violence exposure increases risk for fear-based disorders, such as anxiety, potentially due to disrupted recall of extinguished fear, whereby stimuli previously associated with threat continue to produce a fear response long after they have been deemed safe. However, this emerging work lacks adequate representation of youth from historically marginalized groups, despite their disproportionate exposure to community violence. As such, this study investigates whether such exposure is associated with neurological and behavioral indices of extinction recall in a sample of preadolescent Latina girls. Thirty-five predominantly Mexican-heritage Latina girls (MAge = 10.04, SD = 1.23, range = 8-12 years) underwent functional magnetic resonance imaging (fMRI) while completing an extinction recall task assessing their ability to retrieve related but competing memories of previously conditioned and extinguished threats. Following the fMRI scan, participants self-reported their community violence exposure using the Things I’ve Seen and Heard scale. Relative to low community violence exposed youth, youth exposed to higher levels of community violence demonstrated significantly less ventromedial prefrontal cortex (vmPFC) engagement in response to increasingly ambiguous conditioned stimuli and were also marginally more likely to misattribute threat to conditioned stimuli that were least likely to predict threat. These preliminary results suggest community violence exposure may contribute to disruptions in extinction recall and elucidate a potential mechanism by which these experiences could elevate anxiety, particularly among an understudied group during a sensitive developmental period of heightened anxiety risk.

The physical, biogeochemical, and ecological properties of the modern seafloor are extensively shaped by the activities of burrowing and sediment-mixing animals, processes collectively known as bioturbation. Bioturbation is primarily recorded by homogenized sediments of the seafloor mixed layer and the underlying transition layer of discrete burrows. Although these two zones can be readily measured today, there has been limited understanding of how the mixed and transition layers evolved over the Phanerozoic since animals first began to extensively colonize the seafloor. Here, we provide a record for the depths of the sedimentary mixed and transition layers through the Phanerozoic. We find that although deepening of the sediment mixed layer spanned hundreds of millions of years, a deep transition layer was established as early as the Cambrian and did not further deepen until the Mesozoic-trajectories reflecting evolutionary radiations, changes in nutrient cycling, and alleviation of oxygen stress.

A phenomenological model of two Co adatoms on a Cu(100) substrate was developed using a hybrid approach with “magneto-displacement” effects taken into account. The phenomenological model used parameters obtained from density functional theory (DFT) calculations. The coupled Landau-Lifshitz (LL) equation and Newton’s equation of motion governing magnetic and mechanical degrees of freedom provide the foundation of the model. The modeled to several key results. A mechanism for ferromagnetic (FM)-to-antiferromagnetic (AFM) transition induced by displacement of the adatoms was demonstrated. Mechanical tunability of magnetic resonant frequencies was shown. Displacement-assisted in-plane (IP)-to-out-of-plane (OOP) switching was demonstrated. In addition, the model demonstrates that mechanical displacement can induce an AFM IP-to-FM OOP transition, facilitated by a reduced external magnetic field.