The deployment of Mixture-of-Experts (MoE) models on production high-bandwidth superpods, such as NVIDIA's NVL72/576 and Huawei's CloudMatrix384, introduces critical challenges beyond raw interconnect bandwidth. While these systems provide unified global address spaces and high-bandwidth fabrics, their full potential for sparse MoE communication is hindered by three fundamental bottlenecks: (1) Strict execution serialization imposed by coarse-grained Bulk Synchronous Parallel (BSP) orchestration of interdependent communication phases; (2) Prohibitive synchronization overhead that fails to scale alongside high interconnect bandwidth; and (3) Severe load imbalance resulting from distance-agnostic scheduling of irregular token traffic. To eliminate these bottlenecks, we introduce UBEP (Unified-Bus Expert Parallelism), a production-ready communication library that rethinks MoE's All-to-All primitives for modern superpod architectures. Through large-scale experiments, UBEP reduces All-to-All latency by up to 52.4% and MoE inference Time Per Output Token (TPOT) by up to 11.1%.

Cognitive radar networks have emerged as a key enabler for intelligent unmanned aerial vehicle (UAV) and vehicular sensing, providing adaptive perception, dynamic resource management, and enhanced situational awareness in autonomous and multi-agent environments. Active reconfigurable intelligent surface (ARIS)-assisted cognitive radar systems can provide additional degrees of freedom in signal manipulation and resource allocation, thereby improving multi-platform localization and tracking performance. However, achieving high performance in ARIS-assisted radar systems under limited resources remains a major challenge, as it requires the joint optimization of the ARIS reflection coefficients and transmit power allocation. To address this problem, we propose a reinforcement learning based optimization framework. Specifically, a deep deterministic policy gradient (DDPG) algorithm is employed to optimize the ARIS reflection coefficients, while a multi-agent deep deterministic policy gradient (MADDPG) algorithm is utilized to cooperatively allocate transmit power among multiple radars. Moreover, an immediate reward mechanism and prioritized experience replay scheme are designed to accelerate convergence and stabilize the training process. Simulation results demonstrate that the proposed framework not only enhances system performance but also achieves faster convergence and more efficient resource utilization compared with conventional schemes, then providing a practical and scalable solution for intelligent radar sensing systems.

Unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) has emerged as a valuable supplement to conventional MEC, offering unique advantages in temporary or emergency scenarios. However, the integration of UAVs into MEC introduces new challenges. First, the existing UAV-empowered MEC frameworks predominantly have not accounted for dependencies between the computing tasks. Second, the limited coverage area of UAVs and the sparse distribution of end devices result in a scarcity of training samples, hindering the effectiveness of data-driven approaches. To address these challenges, we introduce a novel UAV-assisted task offloading scheme designed to minimize the average task execution delay and energy consumption in vehicular networks. Initially, we craft a dependency-aware UAV-assisted MEC framework, where the task dependency and priority models are meticulously developed to illustrate the associations between the tasks. Subsequently, we devise an asynchronous federated deep reinforcement learning-based task offloading algorithm, incorporating an asynchronous federated optimization mechanism to enhance the data diversity, with a premise of ensuring data privacy. Our method has been rigorously tested using real traffic flow data and a directed acyclic graph task data set. Comparative experiments highlight the superiority of our framework, showcasing a substantial reduction in the average task delay and energy consumption.

Microsatellite instability (MSI) is a hypermutator phenotype caused by DNA mismatch repair deficiency. MSI has been reported in various human cancers, particularly colorectal, gastric and endometrial cancers. MSI is a promising biomarker for cancer prognosis and immune checkpoint blockade immunotherapy. Several computational methods have been developed for MSI detection using DNA- or RNA-based approaches based on next-generation sequencing. Epigenetic mechanisms, such as DNA methylation, regulate gene expression and play critical roles in the development and progression of cancer. We here developed MSI-XGNN, a new computational framework for predicting MSI status using bulk RNA-sequencing and DNA methylation data. MSI-XGNN is an explainable deep learning model that combines a graph neural network (GNN) model to extract features from the gene-methylation probe network with a CatBoost model to classify MSI status. MSI-XGNN, which requires tumor-only samples, exhibited comparable performance with two well-known methods that require tumor-normal paired sequencing data, MSIsensor and MANTIS and better performance than several other tools. MSI-XGNN also showed good generalizability on independent validation datasets. MSI-XGNN identified six MSI markers consisting of four methylation probes (EPM2AIP1|MLH1:cg14598950, EPM2AIP1|MLH1:cg27331401, LNP1:cg05428436 and TSC22D2:cg15048832) and two genes (RPL22L1 and MSH4) constituting the optimal feature subset. All six markers were significantly associated with beneficial tumor microenvironment characteristics for immunotherapy, such as tumor mutation burden, neoantigens and immune checkpoint molecules such as programmed cell death-1 and cytotoxic T-lymphocyte antigen-4. Overall, our study provides a powerful and explainable deep learning model for predicting MSI status and identifying MSI markers that can potentially be used for clinical MSI evaluation.

Multimodal machine translation (MMT) aims to improve translation quality by equipping the source sentence with its corresponding image.Despite the promising performance, MMT models still suffer the problem of input degradation: models focus more on textual information while visual information is generally overlooked.In this paper, we endeavor to improve MMT performance by increasing visual awareness from an information theoretic perspective.In detail, we decompose the informative visual signals into two parts: source-specific information and target-specific information.We use mutual information to quantify them and propose two methods for objective optimization to better leverage visual signals.Experiments on two datasets demonstrate that our approach can effectively enhance the visual awareness of MMT model and achieve superior results against strong baselines.

Accurate next-day air quality prediction is essential to enable warning and prevention measures for cities and individuals to cope with potential air pollution, such as vehicle restriction, factory shutdown, and limiting outdoor activities. The problem is challenging because air quality is affected by a diverse set of complex factors. There has been prior work on short-term (e.g., next 6 hours) prediction, however, there is limited research on modeling local weather influences or fusing heterogeneous data for next-day air quality prediction. This paper tackles this problem through three key contributions: (1) we leverage multi-source data, especially high-frequency grid-based weather data, to model air pollutant dynamics at station-level; (2) we add convolution operators on grid weather data to capture the impacts of various weather parameters on air pollutant variations; and (3) we automatically group (cross-domain) features based on their correlations, and propose multi-group Encoder-Decoder networks (MGED-Net) to effectively fuse multiple feature groups for next-day air quality prediction. The experiments with real-world data demonstrate the improved prediction performance of MGED-Net over state-of-the-art solutions (4.2% to 9.6% improvement in MAE and 9.2% to 16.4% improvement in RMSE).

This paper describes the model we designed for the Chinese word segmentation Task of NLPCC 2015. We firstly apply a word-based perceptron algorithm to build the base segmenter. Then, we use a Bootstrap Aggregating model of bagging which improves the segmentation results consistently on the three tracks of closed, semi-open and open test. Considering the characteristics of Weibo text, we also perform rule-based adaptation before decoding. Finally, our model achieves F-score 95.12% on closed track, 95.3% on semi-open track and 96.09% on open track.

Click modeling aims to interpret the users' search click data in order to predict their clicking behavior. Existing models can well characterize the position bias of documents and snippets in relation to users' mainstream click behavior. Yet, current advances depict users' search actions only in a general setting by implicitly assuming that all users act in the same way, regardless of the fact that anyone, motivated with some individual interest, is more likely to click on a link than others. It is in light of this that we put forward a novel personalized click model to describe the user-oriented click preferences, which applies and extends matrix / tensor factorization from the view of collaborative filtering to connect users, queries and documents together. Our model serves as a generalized personalization framework that can be incorporated to the previously proposed click models and, in many cases, to their future extensions. Despite the sparsity of search click data, our personalized model demonstrates its advantage over the best click models previously discussed in the Web-search literature, supported by our large-scale experiments on a real dataset. A delightful bonus is the model's ability to gain insights into queries and documents through latent feature vectors, and hence to handle rare and even new query-document pairs much better than previous click models.

Recent advances in click modeling have established it as an attractive approach to interpret search click data. These advances characterize users’ search behavior either in advertisement blocks, or within an organic search block through probabilistic models. Yet, when searching for information on a search result page, one is often interacting with the search engine via an entire page instead of a single block. Consequently, previous works that exclusively modeled user behavior in a single block may sacrifice much useful user behavior information embedded in other blocks. To solve this problem, in this paper, we put forward a novel Whole Page Click (WPC) Model to characterize user behavior in multiple blocks. Specifically, WPC uses a Markov chain to learn the user transition probabilities among different blocks in the whole page. To compare our model with the best alternatives in the Web-Search literature, we run a large-scale experiment on a real dataset and demonstrate the advantage of the WPC model in terms of both the whole page and each block in the page. Especially, we find that WPC can achieve significant gain in interpreting the advertisement data, despite of the sparsity of the advertisement click data.