MOTIVATION: Single-cell technologies enable high-resolution cellular studies but face challenges in identifying differential features due to data complexity. RESULTS: We present OTMODE, a non-parametric method using unbalanced Sinkhorn algorithm and Wald test to improve differential feature identification in single-cell multi-omics data. Under simulation, OTMODE achieved superior performance (average 90% F1 score; average 92% AUC score) with high efficiency (2.2 s for 5000 cells). In practice, it shows greater sensitivity than other state-of-the-art methods in detecting meaningful processes and can evaluate annotation accuracy by identifying potentially misannotated clusters from auto-annotation tools. Furthermore, OTMODE integrates seamlessly with Scanpy, offering a user-friendly solution for researchers. AVAILABILITY AND IMPLEMENTATION: OTMODE is freely available at https://github.com/Eggong/OTMODE and also available at https://pypi.org/project/OTMODE/.

We present DriveGPT, a scalable behavior model for autonomous driving. We model driving as a sequential decision-making task, and learn a transformer model to predict future agent states as tokens in an autoregressive fashion. We scale up our model parameters and training data by multiple orders of magnitude, enabling us to explore the scaling properties in terms of dataset size, model parameters, and compute. We evaluate DriveGPT across different scales in a planning task, through both quantitative metrics and qualitative examples, including closed-loop driving in complex real-world scenarios. In a separate prediction task, DriveGPT outperforms state-of-the-art baselines and exhibits improved performance by pretraining on a large-scale dataset, further validating the benefits of data scaling.

Abstract Background Genome-wide association studies (GWAS) have revealed numerous loci for complex diseases, yet the cellular contexts through which these variants act remain poorly understood. Single-cell RNA sequencing (scRNA-seq) offers high-resolution profiling of cellular heterogeneity, but linking GWAS signals to disease-relevant cells remains a challenge. Methods We developed a statistical framework that integrates GWAS-prioritized genes with single-cell transcriptomes to detect disease-relevant cells. Each target cell is contrasted with pools of healthy reference cells, ensuring that differences reflect disease rather than technical variability. Two controls are implemented: reference cells matched by sequencing depth and expression-matched control genes without known disease associations. Iterative Monte Carlo sampling generates comparisons across random draws of reference cells and control genes, reducing stochastic noise. A difference-of-differences statistic captures disease-specific shifts, with Fisher’s method aggregating significance and enrichment tests identifying cell types overrepresented in patients. Results In simulations and real-world datasets, the framework showed strong performance. In large peripheral blood cohorts, it consistently identified trait-associated populations, with monocytes as a benchmark. Analyses of disease datasets revealed both established and novel pathogenic populations, including age-associated B cells and cytotoxic CD8+ T cells in systemic lupus erythematosus, antigen-presenting acinar cells in Sjögren’s syndrome, and dysregulated proximal tubules in chronic kidney disease. Conclusions These findings demonstrate the utility of integrating GWAS with single-cell transcriptomics to pinpoint disease-relevant cells in complex traits.

Fixed broadband internet service can provide a stable broadband network of up to 100 megabits or even gigabits and users at home can use fixed broadband service for all kinds of internet surfing, including website and application access, watching videos, playing games, etc. Traditional maintenance for fixed broadband networks primarily uses human manual meth-ods, supplemented by some low-level semi-automation operations. Since the long processes with numerous network elements in the fixed broadband network, it is difficult for traditional operation and maintenance to support effectively with high quality. When abnormalities occur, it is quite manpower cost and time cost to monitor and locate faults. Therefore, to improve the autonomous capability of the fixed broadband network, intelligent operation and maintenance methods are necessary. First of all, a brand-new data pre-process method is proposed to detect anomalies and problems of slow access by selecting web services access commonly visited by users. Secondly, as the fixed broadband network is a multi-level and complex structure with only a small amount of anomaly sample data, we propose a multi-source joint anomaly detection model called MSJAD model on multi-dimensional features data. The model validation results on real datasets from the real fixed broadband network are state-of-the-art. The accuracy rate reaches 98 % and the recall is over 99 %. We have already begun to deploy the model on the real fixed broadband network and have achieved good feedback.

Self-supervised learning shows great potential in monocular depth estimation, using image sequences as the only source of supervision. Although people try to use the high-resolution image for depth estimation, the accuracy of prediction has not been significantly improved. In this work, we find the core reason comes from the inaccurate depth estimation in large gradient regions, making the bilinear interpolation error gradually disappear as the resolution increases. To obtain more accurate depth estimation in large gradient regions, it is necessary to obtain high-resolution features with spatial and semantic information. Therefore, we present an improved DepthNet, HR-Depth, with two effective strategies: (1) re-design the skip-connection in DepthNet to get better high-resolution features and (2) propose feature fusion Squeeze-and-Excitation(fSE) module to fuse feature more efficiently. Using Resnet-18 as the encoder, HR-Depth surpasses all previous state-of-the-art(SoTA) methods with the least parameters at both high and low resolution. Moreover, previous SoTA methods are based on fairly complex and deep networks with a mass of parameters which limits their real applications. Thus we also construct a lightweight network which uses MobileNetV3 as encoder. Experiments show that the lightweight network can perform on par with many large models like Monodepth2 at high-resolution with only20%parameters. All codes and models will be available at https://github.com/shawLyu/HR-Depth.

In order to simulate the flux dipping process of microbump flip chip bonding, multiparameter experiments are designed. The dipping process of microbumps is captured by the high-speed camera, so that we can obtain the quantitative relationship between the dipping quantity and its viscosity, dipping speed, dipping depth, and dipping time by virtue of image processing. Furthermore, corresponding mathematical models are established by curve fitting. Finally, all of the numerical models are integrated; numerical simulation is carried out by MATLAB-Simulink and graphical user interface; and a new simulation software is developed, thus providing an important basis and guidance for flux dipping application and parameter selection.

In order to develop the flux dipping process, a quantitative mathematical model that accurately describes the flux dipping process in the flip-chip bonding is proposed. The whole dynamic dipping process is captured by the high-speed transient imaging followed by image processing. Curve fitting of experimental results yields the relation between flux dipping quantity and process parameters. It suggests a quadratic function of quantity (Q) with respect to dipping depth (d), and a piece-wise function to dipping speed (v) including an exponential section and a quadratic or linear section. While the coupled effect of d- and v can be expressed as a merged function of Q = f(d,v), these mathematical models are proved to be effective when applied to the actual dipping process practiced on a flip-chip bonding machine, thus providing a reliable mathematical basis for optimizing dipping flux in industrial manufacturing.

Looking back on the transition of the profiting mode in internal broadcasting and television industry, the paper mainly analyzes the basic framework and applying prospect of a Web living broadcast system based on P2P mode. The system is comprised of four parts: directory server, channel information server, publishing server and client-end program. With the system, traditional TV station media can easily their programs on Internet by configuring one or two servers, which spends just a little cost. At last, we look in a new profiting mode, which is based on the new-generation P2P-IPTV platform for TV stations