The car-following model, which describes how a vehicle follows its leading vehicle within the same lane, is a fundamental component of high-precision microscopic traffic simulation. However, most existing car-following models deal with vehicle interactions without accounting for the influence of macroscopic traffic state on microscopic driving behavior, which makes it difficult to accurately reflect drivers’ behavioral adaptations under varying traffic conditions. To address this limitation, we propose Multiscale Entanglement Theory of Traffic Flow (MET-TF), which explicates the two-way macro-micro causation of the traffic flow system. Based on MET-TF, we further develop the Macro-Micro Coupling Car-Following (MMC-CF) modeling framework, which incorporates both microscopic driving style and macroscopic traffic state as additional input features in a data-driven car-following model for trajectory prediction. To evaluate the effectiveness of the proposed model, extensive experiments with empirical car-following data collected from real-world highway driving scenarios were conducted. The results demonstrate that MMC-CF outperforms baseline models at both microscopic and macroscopic scales. At the microscopic scale, MMC-CF accurately captures the behavior of the following vehicle. At the macroscopic scale, it effectively reconstructs the traffic flow that emerges from the collective behaviors of multiple vehicles, achieving a reduction of at least 52.6% in traffic flow simulation error compared with baseline models. This cross-scale modeling method offers a novel perspective for understanding the microscopic interactions of vehicles and the co-evolutionary dynamics of traffic flow.

In an era overwhelmed by vast amounts of data, the effective curation of web-crawl datasets is essential for optimizing model performance. This paper tackles the challenges associated with the unstructured and heterogeneous nature of such datasets. Traditional heuristic curation methods often inadequately capture complex features, resulting in biases and the exclusion of relevant data. We introduce an advanced, learning-driven approach, Ensemble Curation Of DAta ThroUgh Multimodal Operators, called EcoDatum, which employs a novel quality-guided deduplication method to balance feature distribution. EcoDatum strategically integrates various unimodal and multimodal data curation operators within a weak supervision ensemble framework, utilizing automated optimization to effectively score each data point. EcoDatum, which significantly improves the data curation quality and efficiency, outperforms existing state-of-the-art (SOTA) techniques, ranking 1st on the DataComp leaderboard with an average performance score of 0.182 across 38 diverse evaluation datasets. This represents a 28% improvement over the DataComp baseline method, demonstrating its effectiveness in improving dataset curation and model training efficiency.

Emotional development of children fundamentally relies on secure attachment relationships, yet current AI companions lack the theoretical foundation to provide developmentally appropriate emotional support. We introduce DinoCompanion, the first attachment-theory-grounded multimodal robot for emotionally responsive child-AI interaction. We address three critical challenges in child-AI systems: the absence of developmentally-informed AI architectures, the need to balance engagement with safety, and the lack of standardized evaluation frameworks for attachment-based capabilities. Our contributions include: (i) a multimodal dataset of 128 caregiver-child dyads containing 125,382 annotated clips with paired preference-risk labels, (ii) CARPO (Child-Aware Risk-calibrated Preference Optimization), a novel training objective that maximizes engagement while applying epistemic-uncertainty-weighted risk penalties, and (iii) AttachSecure-Bench, a comprehensive evaluation benchmark covering ten attachment-centric competencies with strong expert consensus. AttachSecure-Bench achieves state-of-the-art performance (57.15%), outperforming GPT-4o and Gemini-2.5-Pro, with exceptional secure base behaviors and superior attachment risk detection. Ablations validate the critical importance of multimodal fusion, uncertainty-aware risk modeling, and hierarchical memory for coherent, emotionally attuned interactions.

With severe challenges facing computer technology scaling, parallel hardware accelerators have become increasingly attractive solutions to continued performance scaling. However, performance scaling is limited for irregular, input-dependent workloads. Often, static analysis is infeasible and overheads can outweigh benefits, resulting sometimes in weaker performance than a serial counterpart.