This paper presents a simple and effective approach to solving the multi-label classification problem. The proposed leverages Transformer decoders query existence of class label. use is rooted in need extracting local discriminative features adaptively for different labels, which strongly desired property due multiple objects one image. built-in cross-attention module decoder offers an way label embeddings as queries probe pool class-related from feature map computed by vision backbone...

Accurately building height estimation from remote sensing imagery is an important and challenging task. However, the existing shadow-based methods have large errors due to complex environment in imagery. In this paper, we propose a multi-scene method based on shadow high resolution First, of classified described by analyzing features Second, variety models established different scenes. addition, regularization extraction proposed, which can solve problem mutual adhesion shadows dense areas...

A key benefit of Amazon EC2-style cloud computing service is the ability to instantiate a large number virtual machines (VMs) on fly during flash crowd events. Most existing research focuses policy decision such as when and where start VM for an application. In this paper, we study different problem: how can VMs applications inside be brought up quickly possible? This problem has not been solved satisfactorily in services. We develop fast technique by restoring previously created snapshots...

Beach surface moisture (BSM) is crucial to studying coastal aeolian sand transport processes. However, traditional measurement techniques fail accurately monitor distribution with high spatiotemporal resolution. Remote sensing technologies have garnered widespread attention for providing rapid and non-contact measurements, but a single method has inherent limitations. Passive remote challenged by complex beach illumination sediment grain size variability. Active represented LiDAR (light...

Challenges are emerging in testing service-oriented architecture (SOA) systems. Current is not sufficient to deal with the new requirements arising from several SOA features such as composition, loose coupling, and code without a graphical user interface. The most critical information of an solution actually how services composed interact each other. This paper proposes gray-box approach, that is, approach involves having access internal workings, data structures, algorithms when designing...

It is very attractive to formulate vision in terms of pattern theory \cite{Mumford2010pattern}, where patterns are defined hierarchically by compositions elementary building blocks. But applying real world images currently less successful than discriminative methods such as deep networks. Deep networks, however, black-boxes which hard interpret and can easily be fooled adding occluding objects. natural wonder whether better understanding networks we extract blocks used develop theoretic...

The quality of service oriented architecture (SOA) solutions is becoming more and important along with the increasing adoption SOA. Continuous integration testing (CIT) an effective technology to discover bugs as early possible. However, diversity programming models used in SOA solution distribution nature pose new challenges for CIT. Existing frameworks focus on applications developed by a single model. In this paper, unified test framework proposed overcome these limitations enable CIT...

Federated reinforcement learning aims to promote training efficiency or improve policy quality through information interaction with privacy protection. Existing federated methods rarely utilize the structure of algorithms while are limiting specific scenarios algorithms. We propose a general framework FRS, which employs reward shaping as shared among different clients tasks each client's speed and quality. The is implicitly learned by average state value all protect task real trajectory...

With the increasing application of deep learning in various domains, salient object detection optical remote sensing images (ORSI-SOD) has attracted significant attention. However, most existing ORSI-SOD methods predominantly rely on local information from low-level features to infer boundary cues and supervise them using ground truth, but fail sufficiently optimize protect information, almost all approaches ignore potential advantages offered by last layer decoder maintain integrity...

Selective regression testing involves retesting of software systems with a subset the test suite to verify that modifications have not adversely impacted existing functions. Although this problem has been heavily researched, it never discussed in context SaaS (Software as service). This paper presents specific requirements, challenges and benefits delivering selection service (RTaaS). We will introduce how design implement RTaaS platform. An implementation piloted improved via several real...

Deep convolutional neural networks (CNNs) have achieved breakthrough performance in many pattern recognition tasks such as image classification. However, the development of high-quality deep models typically relies on a substantial amount trial-and-error, there is still no clear understanding when and why model works. In this paper, we present visual analytics approach for better understanding, diagnosing, refining CNNs. We formulate CNN directed acyclic graph. Based formulation, hybrid...

We propose a unified game-theoretical framework to perform classification and conditional image generation given limited supervision. It is formulated as three-player minimax game consisting of generator, classifier discriminator, therefore referred Triple Generative Adversarial Network (Triple-GAN). The generator the characterize distributions between images labels classification, respectively. discriminator solely focuses on identifying fake image-label pairs. Under nonparametric...

Oriented object detection has recently attracted increasing attention for its importance in aerial image processing. Popular methods oriented and densely packed objects usually utilize the rotation angle to reduce overlap of bounding boxes over horizontal line. However, those angle-based remain challenging due angular periodicity regression inconsistent, which are uniformly summarized as uncertainty. Previous heal uncertainty problem by adding complex limits on loss function. In this paper,...

This paper presents ORTS, a tool for facilitating testers to generate optimized regression test suite commercial Java applications. The aspects emphasized by the demonstration are: (1) how help capture runtime traces of execution; (2) identify change points during build update; (3) does ORTS improve efficiency testing and reduce cost generating suite. whole design strategy is lightweight, making selection process more automated effective, scalable scenarios with resource time constraints.

Selective regression testing involves re-testing of software systems with a subset the whole test suite to verify that modifications have not caused adverse impacts existing functions complied requirements specifications. With growing globalization and individual services providers, many development teams belong different organizations, often only get binary release application without access its source code. This makes code analysis based selection strategy applicable. Meanwhile approach...

Automatically writing stylized Chinese characters is an attractive yet challenging task due to its wide applicabilities. In this paper, we propose a novel framework named Style-Aware Variational Auto-Encoder (SA-VAE) flexibly generate characters. Specifically, capture the different characteristics of character by disentangling latent features into content-related and style-related components. Considering complex shapes structures, incorporate structure information as prior knowledge our...

Selective regression testing involves retesting of software systems with a subset the test suites to verify that modifications have not adversely impacted existing functions. Although this problem has been heavily researched, it never discussed in context SaaS (Software as service). This paper presents specific requirements, challenges and benefits delivering selection service (RTaaS). We will introduce how design implement RTaaS platform. An implementation piloted improved via several real...

Deep generative models have shown promising results in generating realistic images, but it is still non-trivial to generate images with complicated structures. The main reason that most of the current fail explore structures including spatial layout and semantic relations between objects. To address this issue, we propose a novel deep structured model which boosts adversarial networks (GANs) aid structure information. In particular, or scene encoded by stochastic and-or graph (sAOG),...

Individual trajectory traces of different lengths often amount to hundreds or thousands points distributed over continuous spatial space. This makes fast pattern mining very challenging. For road network constrained trajectories like vehicle trajectories, mapping raw links is a natural calibration procedure that can greatly alleviate the complexity subsequent mining. However, map generally proprietary and imposes limitations on commercial applications. Although variety inference approaches...