Annals of the New York Academy of Sciences最新文献

The model minority myth casts Asians as seamlessly integrated into science, technology, engineering, and mathematics (STEM) fields through merit alone. This narrative often leads researchers to treat Asians as statistically similar to Whites or as a homogeneous group. Our study challenges this framing by examining how race is experienced within STEM mentoring relationships between undergraduates and their faculty or postgraduate mentors. We employ a race-intentional methodology with three components: (1) centering race consciousness in mentoring as a key indicator of racialized experience; (2) disaggregating Asian and White experiences; and (3) analyzing how race intersects with gender and power positioning (mentors vs. mentees). Using survey data from 709 participants in STEM research programs, we find that Asians generally report greater race consciousness than their White counterparts, with particularly heightened awareness among Asian women and nonbinary individuals. Asian undergraduate mentees report significantly more race consciousness than White mentees, but this disparity disappears at the mentor level. These findings expose critical flaws in both the meritocratic ideals of STEM and the model minority narrative—especially at a time when Asian achievement is increasingly being politically and legally weaponized to promote colorblind ideologies.

Autonomous forklift operations in modern warehouses face critical challenges from dense pallet stacking, partial occlusions, and variable illumination conditions, which significantly impact positioning accuracy and operational efficiency. This paper presents a novel hypergraph computing and knowledge-enhanced framework for robust pallet pose estimation, addressing these complex industrial scenarios. The framework leverages hypergraph structures to model high-order spatial relationships among pallet keypoints, effectively representing geometric dependencies in E-shaped pallet configurations even under severe occlusion and stacking artifacts. By integrating hypergraph computing with domain knowledge, we develop a hyper-pose architecture that combines hierarchical attention fusion mechanisms with geometry-aware keypoint detection modules, encoding pallet structural priors as differentiable constraints. A topology-preserving pruning strategy specifically designed for hypergraph structures reduces the model to 20.2 MB, while maintaining 97.5% detection accuracy and 97.3% keypoint localization precision. The framework incorporates uncertainty constrained pose estimation using Mahalanobis distance optimization, achieving angular errors below 1.6° and distance errors under 18 mm. The system demonstrates real-time performance at 72.1 FPS on the NVIDIA Jetson Orin Nano edge computing platform. Extensive validation using 16,233 warehouse images confirms robust performance across challenging conditions. The integration of hypergraph computing with knowledge enhancement establishes a new paradigm for industrial vision systems, significantly improving automated material handling reliability.

Spiking neural networks (SNNs) possess great potential for energy-efficient computation; however, their practical deployment is often constrained by the high computational cost associated with training across multiple time steps. Unlike previous approaches that simply reduce the number of time steps, this work investigates temporal feature redundancy and reveals the feature overlapping phenomenon in SNNs—showing that substantial computational redundancy exists across temporal dimensions. To address this issue, we propose our core contribution, temporal differential decoupling (TDD), which transforms network computation into the differential domain to disentangle static and dynamic feature components. This decoupling enables focused processing of informative signals and significantly reduces redundant computation without compromising model accuracy. Building upon this framework, we further design the TDD-based differential domain low-sparsity approximation (TDD-DDLA) algorithm, which is based on the gradient sensitivity criterion, as an implementation strategy to quantify each temporal feature's contribution to gradient updates and achieve efficient energy optimization. In contrast to prior studies that only adjusted time steps without explicit feature-level analysis, our framework provides a structured and theoretically grounded analysis of temporal feature evolution. Experimental results demonstrate that our method achieves up to 80.9% fewer spikes per time step and 57.8% fewer total spikes, with no degradation in classification performance, offering a promising pathway toward scalable, low-cost, and high-accuracy SNN deployment.

Helicobacter mastomyrinus (H.m), a Helicobacter species colonizing rodent liver and intestine, produces cytolethal distending toxin (CDT), which induces host cell damage. While the active CdtB subunit is linked to various diseases, the pathogenic mechanisms of H.m in the liver and the precise actions of CdtB remain incompletely defined. This study generated a CdtB-deficient strain (H.m ΔCdtB) and compared it with the wild-type H.m (H.m WT). Male BALB/c mice were orally infected via gavage. The impacts of CdtB on liver injury, inflammation, oxidative stress, endoplasmic reticulum (ER) stress, and calcium homeostasis were systematically evaluated. Infection with H.m for 12 and 24 weeks induced significant hepatic inflammation and necrosis and a significant increase in hepatocyte proliferation. CdtB deficiency did not impair H.m colonization but markedly reduced inflammatory severity. CdtB facilitated the secretion of proinflammatory cytokines (IL-1β, IL-6, MCP-1, TNF-α), driving chronic liver inflammation. Mechanistically, H.m infection upregulated ER stress-related genes and proteins, increased calcium ion levels, whereas H.m ΔCdtB infection showed no significant effects. Notably, pharmacological inhibition of ER stress by 4-phenylbutyric acid alleviated calcium imbalance and attenuated liver injury. These findings indicate that H.m CdtB induces ER stress, disrupting intracellular calcium homeostasis and exacerbating liver injury in male BALB/c mice.

Intracerebral hemorrhage (ICH) causes severe brain damage, with perihematomal edema (PHE) contributing to poor outcomes. However, the biomechanical properties of PHE tissue remain incompletely characterized. In this study, we examined poroviscoelastic behavior in edematous brain tissue using a rat ICH model, together with associated structural and molecular alterations. Load-relaxation tests were used to characterize viscoelastic and poroelastic responses, while MRI (magnetic resonance imaging) and histological analyses were performed to assess edema progression and tissue integrity. Proteomics was conducted to characterize molecular features associated with PHE tissue. The results showed that PHE tissue exhibited cellular edema, varying power-law exponents, and altered effective diffusion coefficients. Histological analysis revealed disrupted parenchymal cell networks and reduced extracellular space. Proteomics identified changes in the abundance of cell adhesion proteins and extracellular matrix components, including chondroitin sulfate proteoglycans. Together, these findings provide an integrated description of mechanical, structural, and molecular alterations in perihematomal tissue following ICH, offering a biomechanical perspective on tissue changes during PHE progression.