Growing main memory capacity has fueled the development of in-memory big data management and processing. By eliminating disk I/O bottleneck, it is now possible to support interactive analytics. However, systems are much more sensitive other sources overhead that do not matter in traditional I/O-bounded disk-based systems. Some issues such as fault-tolerance consistency also challenging handle environment. We witnessing a revolution design database exploits its storage layer. Many these...

BACKGROUND Early screening methods for gastric cancer (GC) are lacking; therefore, the disease often progresses to an advanced stage when patients first start exhibit typical symptoms. Endoscopy and pathological biopsy remain primary diagnostic approaches, but they invasive not yet widely applicable early population screening. miRNA is a highly conserved type of RNA that exists stably in plasma. Dysfunction linked tumorigenesis progression, indicating individual miRNAs or combinations...

Semi-supervised clustering is an very important topic in machine learning and computer vision. The key challenge of this problem how to learn a metric, such that the instances sharing same label are more likely close each other on embedded space. However, little attention has been paid better representations when data lie non-linear manifold. Fortunately, deep led great success feature recently. Inspired by advances learning, we propose transductive semi-supervised maximum margin approach....

Multimodal learning with deep Boltzmann machines (DBMs) is an generative approach to fuse multimodal inputs, and can learn the shared representation via Contrastive Divergence (CD) for classification information retrieval tasks. However, it a 2-fan DBM model, cannot effectively handle multiple prediction Moreover, this model recover hidden representations well by sampling from conditional distribution when more than one modalities are missing. In paper, we propose K-fan structure which...

目的 海洋涡旋精准检测是揭示海洋涡旋演变规律及其与其他海洋现象相互作用的基础。然而，海洋涡旋在其活跃海域呈现小尺度目标、密集分布的特点，导致显著的检测精度低问题。传统方法受限于人工设计参数缺乏泛化能力，而深度学习模型的高采样率在检测小目标过程中底层细节和轮廓等信息损失严重，使得目标检测轮廓与目标真实轮廓相差甚远。针对海洋涡旋小目标特点导致检测精度低，高采样率深度模型检测轮廓不精确的问题，提出一种改进的U-Net网络。方法 该模型基于渐进式采样结构，为获取上下文信息提升不同极性海洋涡旋目标的检测精度，增加上下文特征融合模块；为增加该模块对海洋涡旋小目标的关注，在特征融合前对最底层特征嵌入残差注意力模块，使模型可以更多关注海洋涡旋的轮廓信息。最后引入数据扩充方法缓解模型存在的过拟合问题。结果 本文以南大西洋的卫星海表面高度数据集开展实验，结果表明，本文模型检测准确率达到了93.24%，同时在海洋涡旋的检测数量上与真实结果更加接近，验证了模型在小目标检测方面的性能更加优秀。结论...

Digitized pathological diagnosis has been in increasing demand recently. It is well known that color information critical to the automatic and visual analysis of slides. However, variations due various factors not only have negative impact on pathologist's diagnosis, but also will reduce robustness algorithms. The cause differences are process making slices, digitization. Different strategies proposed alleviate variations. Most such techniques rely collecting statistics perform matching...

Restricted Boltzmann machines (RBMs) and their variants have attracted a lot of attention recently. They been applied widely, e.g., In handwriting recognition, document categorization object recognition. Unfortunately, an RBM requires large parameter space since it is fully-connected bipartite graph, especially with high dimensional input spaces. Moreover, still unclear how selects discriminative features for classification problems. This necessitates the selection effective this paper, we...

In this paper, we propose an enhanced end-to-end object detector based on Sparse R-CNN (EnSparse R-CNN), which aims at backbone, neck and head of detector. For the backbone network, constructure aggregate residual network (ARNet) with connection, can achieve global features to improve ability feature extraction by extracting salient from maps. build attention context pyramid (ACFPN) fuse features, integrate multi-layer make full use features. detection head, design a gradually refined...

Programmable logic controllers (PLCs) are widely used in computer-based industrial applications. Timers play a pivotal role PLC real-time embedded system The paper addresses the formal validation of systems with timers theorem proving Coq. timer behavior is characterized formally. A refinement methodology presented terms an abstract model and concrete model. calibrated by mapping relation. soundness shown system. An illustrative case study demonstrates effectiveness approach.

High-energy physics computing is a typical data-intensive calculation. Each year, petabytes of data needs to be analyzed, and access performance increasingly demanding. The tiered storage system scheme for building unified namespace has been widely adopted. Generally, stored on devices with different performances prices according frequency. When the heat changes, then migrated appropriate tier. At present, heuristic algorithms based artificial experience are used in prediction. Due...

Text style transfer is one of the controllable text generation tasks, which can convert attributes. The mainstream method to separate content and text, then combine vectors with other generate. However, implicit expression cannot completely meaning style, separation recombination may also lead a decrease in naturalness fluency text. Therefore, we propose new idea, first encodes into latent representation, iteratively optimizes representation smallest changes achieve transfer. Introducing...

Diffusion probabilistic models have generated high quality image synthesis recently. However, one pain point is the notorious inference to gradually obtain clear images with thousands of steps, which time consuming compared other generative models. In this paper, we present a shortcut MCMC sampling algorithm, balances training and inference, while keeping data's quality. particular, add global fidelity constraint combat local fitting from diffusion We do some initial experiments show very...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI