This paper studies the problem of automatic detection false rumors on Sina Weibo, popular Chinese microblogging social network. Traditional feature-based approaches extract features from rumor message, its author, as well statistics responses to form a flat feature vector. ignores propagation structure messages and has not achieved very good results. We propose graph-kernel based hybrid SVM classifier which captures high-order patterns in addition semantic such topics sentiments. The new...

With the high spectral resolution, hyperspectral image (HSI) can provide a wealth of information for classification. Many classification methods utilize training samples to classify ground materials. However, small sample problem is still urgent be solved when considering cost labeling samples. In order solve this problem, letter proposes semisupervised method based on simple linear iterative cluster (SLIC) segmentation HSI. This improves SLIC better explore characteristic It explores...

Due to the relatively low temporal resolutions of high spatial resolution (HR) remotely sensed images, land-cover change detection (LCCD) may have use multi-temporal images with different resolutions. The (LR) often repetition rates, but they contain a large number mixed pixels, which seriously limit their capability in detection. Soft classification (SC) can produce proportional fractions land-covers, on sub-pixel mapping (SPM) construct fine maps reduce low-spatial-resolution-problem some...

In recent decades, lithological mapping techniques using hyperspectral remotely sensed imagery have developed rapidly. The processing chains visible-near infrared (VNIR) and shortwave (SWIR) data are proven to be available in practice. thermal (TIR) portion of the electromagnetic spectrum has considerable potential for mineral lithology mapping. particular, abovementioned rocks at wavelengths 8–12 μm were found discriminative, which can seen as a characteristic apply classification....

Hyperspectral images contain abundant spectral information, which provide great potential for target detection. However, it also introduces a critical variability problem hyperspectral detection, makes the detection much difficult than classical match issue. Many traditional methods have been proposed to deal with variability. these algorithms are still highly susceptible The single input restriction and inherent characteristics mining main issues methods. multitask learning (MTL) technique...

Deep learning has achieved good performance in hyperspectral image classification (HSIC). Many methods based on deep use and complex network structures to extract rich spectral spatial features of images (HSIs) with high accuracy. During the process, how accurately information from pixel blocks HSIs is important. All are treated equally classification, input often contains much useless information, leading a low result. To solve this problem, an enhanced spectral-spatial residual attention...

With the development of hyperspectral sensors, availability images (HSIs) has increased significantly, prompting advancements in deep learning-based image classification (HSIC) methods. Recently, graph convolutional networks (GCNs) have been proposed to process graph-structured data non-Euclidean domains, and used for HSIC. The superpixel segmentation should be implemented first GCN-based methods, however, it is difficult manually select optimal sizes obtain useful information...

Target detection from hyperspectral images is an important problem but encounters a critical challenge of simultaneously reducing spectral redundancy and preserving the discriminative information. Recently, joint sparse representation multi-task learning (JSR-MTL) approach was proposed to address challenge. However, it does not fully explore prior class label information training samples difference between target dictionary background when constructing model. Besides, there may exist...

Extracting subpixel land-cover change detection (SLCCD) information is important when multitemporal remotely sensed images with different resolutions are available. The general steps as follows. First, soft classification applied to a low-resolution (LR) image generate the proportion of each class. Second, differences produced by use another high-resolution (HR) and used input mapping. Finally, sharpened difference map can be generated. However, prior HR only compare enhanced LR for...

We present algorithms, implemented as an extension to the OpenFst library, that yield a class of transducers encode linear models for structured inference tasks like segmentation and tagging.This allows use general finite-state operations with such models.For instance, composition can be used apply model lattice input (or other more automata) then result automaton passed subsequent processing shortest path algorithms.We demonstrate library on graphemeto-phoneme conversion, encoding multiple...

Net primary productivity (NPP) is a critical component in terrestrial ecosystem carbon cycles. Thus, quantitatively estimating and monitoring the dynamics of NPP have become key aspects for exploring cycle ecosystems. Anthropogenic activity, such as urbanization, has significant effects on increases pressure natural resources specific region. However, to date, although many studies focused relationship between variation they usually ignored any differences at long-term spatiotemporal...

Hyperspectral anomaly detection involves in many practical applications. Traditional methods are mainly proposed based on statistical models and geometrical models. This paper proposes an Otsu-based isolation forest method, which applies the assumption that pixels more sensitive to be isolated from alternative pixels. The trains by assembling multiple binary trees. To construct a discriminative tree, splitting criterion is applied split subsamples into two groups at each division. Then, it...

Hyperspectral remote sensing image (HSI) include rich spectral information that can be very beneficial for change detection (CD) technology. Due to the existence of many mixed pixels, pixel-wise approaches lead considerable errors in resulting CD map. The unmixing (SU) method is a potential solution this problem, as it decomposes pixels into set fractions land cover. Subsequently, map created by comparing abundance images. However, based only on images through SU method, they are unable...

Hyperspectral imagery contains a large number of mixed pixels, which limits its utility. Super-resolution mapping is potential solution to this problem, designed use the proportion land covers obtain sharpened thematic map with higher resolution. Endmember fundamental variable in process, critical issue for decomposing pixels and sharpening subpixel level images. In most cases, forms endmember combination diverse are very distinct. However, traditional soft classification methods neglect...

Due to the existence of mixed pixels in a remote sensed image, traditional change detection (CD) methods at "full-pixel level" are often unable provide detailed changed information effectively. A subpixel (SCD) technique can deal with this issue two steps: soft classification is applied derive proportional differences from coarse multitemporal images, and then sharpened thematic map fine spatial resolution generated based on mapping. However, changes endmember combination within ignored,...

Target detection is playing an important role in hyperspectral image (HSI) processing. Many traditional methods utilize the discriminative information within all single-band images to distinguish target and background. The critical challenge with these simultaneously reducing spectral redundancy preserving information. multitask learning (MTL) technique has potential solve aforementioned challenge, since it can further explore inherent similarity between adjacent images. This letter proposes...

High-resolution and thermal infrared hyperspectral data acquired from the Thermal Infrared Airborne Spectrographic Imager (TASI) have been recognized as efficient tools in geology, demonstrating significant potential for rock discernment. Deep learning (DL), an advanced technology, has driven substantial advancements lithological mapping by automatically extracting high-level semantic features images to enhance recognition accuracy. However, gathering sufficient high-quality samples model...