Federated learning is a distributed training method that integrates multi-party data information using privacy-preserving technologies through dispersed client data sets to jointly construct a global model under the c...
详细信息
Recent phenomena such as pandemics,geopolitical tensions,and climate change-induced extreme weather events have caused transportation network interruptions,revealing vulnerabilities in the global supply chain.A salien...
详细信息
Recent phenomena such as pandemics,geopolitical tensions,and climate change-induced extreme weather events have caused transportation network interruptions,revealing vulnerabilities in the global supply chain.A salient example is the March 2021 Suez Canal blockage,which delayed 432 vessels carrying cargo valued at$92.7 billion,triggering widespread supply chain *** ability to model the spatiotemporal ramifications of such incidents remains *** fill this gap,we develop an agent-based complex network model integrated with frequently updated maritime *** Suez Canal blockage is taken as a case *** results indicate that the effects of such blockages go beyond the directly affected countries and *** Suez Canal blockage led to global losses of about$136.9($127.5–$147.3)billion,with India suffering 75%of these *** losses show a nonlinear relationship with the duration of blockage and exhibit intricate trends post *** proposed model can be applied to diverse blockage scenarios,potentially acting as an earlyalert system for the ensuing supply chain ***,high-resolution daily data post blockage offer valuable insights that can help nations and industries enhance their resilience against similar future events.
As the size of datasets and neural network models increases, automatic parallelization methods for models have become a research hotspot in recent years. The existing auto-parallel methods based on machine learning or...
详细信息
As the size of datasets and neural network models increases, automatic parallelization methods for models have become a research hotspot in recent years. The existing auto-parallel methods based on machine learning or...
As the size of datasets and neural network models increases, automatic parallelization methods for models have become a research hotspot in recent years. The existing auto-parallel methods based on machine learning or graph algorithms still have issues with search efficiency and applicability. This paper proposes an automatic parallel method based on a dual-population genetic algorithm, TGA, which transforms model partitioning and placement into an integer linear programming problem and constructs a cost model to evaluate the solution. The solution space is built using the neural network’s dataflow graph and device cluster’s topology, and the dual-population genetic algorithm is used to search for the optimal model parallel strategy. Experiments with various models show that the proposed method can improve single-step execution time by up to 42% compared to the Baechi method and up to 37.7% compared to the Hierarchical method.
Dear editor,Software developers tend to reuse existing libraries to facilitate their development process and implement certain functionalities by invoking application programming interfaces(APIs) [1]. However, it rema...
详细信息
Dear editor,Software developers tend to reuse existing libraries to facilitate their development process and implement certain functionalities by invoking application programming interfaces(APIs) [1]. However, it remains a challenging task for developers to correctly use APIs [2], so they often consult API learning resources [3, 4]. As one of the most important API learning resources,
Deep neural networks (DNNs) have achieved remarkable performance across a wide range of applications, while they are vulnerable to adversarial examples, which motivates the evaluation and benchmark of model robustness...
详细信息
In recent years, the growing demand for medical imaging diagnosis has placed a significant burden on radiologists. As a solution, Medical Vision-Language Pre-training (Med-VLP) methods have been proposed to learn univ...
详细信息
With the increasing of software complexity and user demands, collaborative service is becoming more and more popular. Each service focuses on its own specialty, their cooperation can support complicated task with high...
详细信息
Visual Question Answering (VQA) requires a finegrained and simultaneous understanding of both the visual content of images and the textual content of questions. Therefore, designing an effective 'co-attention'...
详细信息
ISBN:
(纸本)9781728132945
Visual Question Answering (VQA) requires a finegrained and simultaneous understanding of both the visual content of images and the textual content of questions. Therefore, designing an effective 'co-attention' model to associate key words in questions with key objects in images is central to VQA performance. So far, most successful attempts at co-attention learning have been achieved by using shallow models, and deep co-attention models show little improvement over their shallow counterparts. In this paper, we propose a deep Modular Co-Attention Network (MCAN) that consists of Modular Co-Attention (MCA) layers cascaded in depth. Each MCA layer models the self-attention of questions and images, as well as the question-guided-attention of images jointly using a modular composition of two basic attention units. We quantitatively and qualitatively evaluate MCAN on the benchmark VQA-v2 dataset and conduct extensive ablation studies to explore the reasons behind MCAN's effectiveness. Experimental results demonstrate that MCAN significantly outperforms the previous state-of-the-art. Our best single model delivers 70.63% overall accuracy on the test-dev set.
Recently, deep learning is widely developed in computer vision applications. In this paper, a novel simple tracker with deep learning is proposed to complete the tracking task. A simple fully convolutional Siamese net...
详细信息
暂无评论