The estimation of conditional quantiles at extreme tails is of great interest in numerous applications. Various methods that integrate regression analysis with an extrapolation strategy derived from extreme value theory have been proposed to estimate extreme conditional quantiles in scenarios with a fixed number of covariates. However, these methods prove ineffective in high-dimensional settings, where the number of covariates increases with the sample size. In this article, we develop new estimation methods tailored for extreme conditional quantiles with high-dimensional covariates. We establish the asymptotic properties of the proposed estimators and demonstrate their superior performance through simulation studies, particularly in scenarios of growing dimension and high dimension where existing methods may fail. Furthermore, the analysis of auto insurance data validates the efficacy of our methods in estimating extreme conditional insurance claims and selecting important variables.
Introduction Sex differences in cardiovascular disease (CVD) risk are examined through biological and clinical factors, with less attention to early-life social exposures. This study examined associations of childhood parental son preference with CVD risk, sex differences, and mediation by modifiable risk factors. Methods This cohort analysis included China Health and Retirement Longitudinal Study participants aged ≥45 years without baseline CVD. Parental son preference was assessed retrospectively in 2014; incident CVD was self-reported physician-diagnosed heart disease or stroke through 2020. Sampling-weighted, community-clustered Cox models estimated adjusted hazard ratios (aHRs) and 95% CIs. Sex was prespecified as an effect modifier; mediation by 13 risk factors used inverse-odds-ratio weighting. Data were collected from 2011 to 2020 and analyzed from 2025 to 2026. Results Among 8,079 participants (mean age, 57.5 years; 4,216 women [52.2%]), 1,820 (22.5%) reported parental son preference. Son preference was associated with higher CVD risk overall (aHR 1.23 [95% CI 1.04, 1.46]) and among women (aHR 1.25 [95% CI 1.03, 1.53]); among men, the estimate was 1.16 (95% CI 0.87, 1.56), with limited heterogeneity by sex (ratio of aHRs 1.06 [95% CI 0.72, 1.58]). Among women, risk was concentrated in the highest paternal (aHR 1.47 [95% CI 1.14, 1.90]) and maternal (aHR 1.50 [95% CI 1.12, 1.99]) preference categories. Modifiable risk factors mediated 5.8% (95% CI 1.9%, 9.7%) of the association among women, mainly through socioeconomic and psychosocial factors. CVD risk was highest with both son preference and high risk-factor burden overall (aHR 1.97 [95% CI 1.43, 2.73]) and among women (aHR 2.23 [95% CI 1.61, 3.10]). Conclusions Parental son preference was associated with higher incident CVD risk, with the largest estimates in the highest paternal or maternal categories among women. Modifiable risk factors explained a modest proportion, supporting life-course cardiovascular prevention that considers sex-differentiated childhood environments alongside risk-factor modification.
Wasserstein autoregression provides a robust framework for modeling serial dependence among probability distributions, with wide-ranging applications in economics, finance, and climate science. In this paper, we develop portmanteau-type diagnostic tests for assessing the adequacy of Wasserstein autoregressive models. By defining autocorrelation functions for model errors and residuals in the Wasserstein space, we construct two related tests: one analogous to the classical McLeod type test, and the other based on the sample-splitting approach of Davis and Fernandes(2025). We establish that, under mild regularity conditions, the corresponding test statistics converge in distribution to chi-square limits. Simulation studies and empirical applications demonstrate that the proposed tests effectively detect model mis-specification, offering a principled and reliable diagnostic tool for distributional time series analysis.
Modern serving systems for Mixture-of-Experts (MoE) models adopt hybrid data-expert parallelism: expert parallelism (EP) shards experts across GPUs to scale capacity, while data parallelism (DP) replicates attention layers across instances to process independent requests. Existing systems bind each request's attention, MoE communication, and KV cache to a single instance. Because attention latency scales with KV cache size while MoE communication latency scales with batch size, this binding cannot balance both simultaneously, producing EP stragglers; it also fragments KV memory across instances, inflating tail latency under long contexts. While existing context parallelism (CP) mitigates these constraints, existing group-level and uniform CP designs incur prohibitive communication and attention-side overheads. We present NanoCP, which decouples MoE communication from KV cache placement and achieves dual balance through dynamic context parallelism (DCP). DCP assigns each request a context-parallel degree sized to its KV footprint: long requests distribute attention across multiple instances; short requests remain local. This dynamic parallelism effectively liquefies the KV cache across DP instances within each node, balancing both the per-instance KV cache occupancy and batch sizes without unnecessary load-balancing costs. To bridge DCP with static execution, NanoCP introduces an ahead-of-time (AOT) graph engine paired with a custom routing-based communication backend. Experimental results show that, under 99
Text-Based Person Search (TBPS) aims to retrieve target person images from a large-scale database using natural language descriptions, serving as a critical task in multimodal perception and visual pattern recognition. Bridging the semantic gap between heterogeneous modalities while capturing fine-grained correspondences remains a fundamental challenge, especially when discriminating visually similar individuals based on complex textual semantics. To address these challenges, we propose Sew Calibration and Masked Modeling (SCMM), a unified framework that calibrates cross-modal representations for effective multimodal visual-textual pattern matching. Concretely, SCMM introduces two principal components: a sew calibration loss that dynamically aligns image-text features via a quality-guided adaptive margin governed by textual information density, and a masked caption modeling loss that establishes fine-grained semantic correspondences through transformer-based masked prediction. The sew calibration mechanism imposes bidirectional constraints to compactly cluster same-identity features in a shared embedding space. Simultaneously, the masked modeling component acts as a cross-modal decoder that learns word-level representations, effectively discriminating subtle attribute differences. Importantly, our dual-encoder architecture strikes an optimal balance between representation expressiveness and computational efficiency by adopting a training-only decoder design. Extensive experiments on CUHK-PEDES, ICFG-PEDES, and RSTPReID datasets demonstrate that SCMM achieves state-of-the-art performance with Rank-1 accuracies of 73.81%, 64.25%, and 57.35%, respectively. Thorough ablation studies confirm the efficacy of each proposed mechanism in establishing robust cross-modal patterns for multimodal perception and recognition.
