Histopathological image classification (HIC) and content-based histopathological retrieval (CBHIR) are two promising applications for the whole slide (WSI) analysis. HIC can efficiently predict type of lesion involved in a image. In general, aid pathologists locating high-risk cancer regions from WSI by providing cancerous probability map WSI. contrast, CBHIR was developed to allow searches with similar content region interest (ROI) database consisting historical cases. Sets cases accessible...

Abstract. Using an innovative process model, we describe and analyse the of introducing enterprise resource planning (ERP) systems in two Chinese small medium‐sized enterprises especially their decisions concerning business re‐engineering. First compared results from our cases with Martinsons' earlier work (2004). One case seemed to fit most characteristics a private venture (PV) whereas other case, also PV, had very low degree fit. We used model offer further insights features such as its...

Knowledge graph (KG) entity typing aims at inferring possible missing type instances in KG, which is a very significant but still under-explored subtask of knowledge completion. In this paper, we propose novel approach for KG trained by jointly utilizing local from existing assertions and global triple KGs. Specifically, present two distinct knowledge-driven effective mechanisms inference. Accordingly, build embedding models to realize the mechanisms. Afterward, joint model via connecting...

The latest evolution of wireless communications enables user access rich Virtual Reality (VR) services via the Internet, including while on move. However, providing a premium immersive experience for massive number concurrent users with various device configurations is significant challenge due to ultra-high data rate and ultra-low delay requirements live VR services. This paper introduces an innovative multi-user cost-efficient crowd-assisted delivery computing (MEC-DC) framework, which...

Stock Movement Prediction (SMP) aims at predicting listed companies' stock future price trend, which is a challenging task due to the volatile nature of financial markets. Recent studies show that momentum spillover effect plays significant role in fluctuation. However, previous typically only learn simple connection information among related companies, inevitably fail model complex relations companies real market. To address this issue, we first construct more comprehensive Market Knowledge...

In the field of pathology, whole slide image (WSI) has become major carrier visual and diagnostic information. Content-based retrieval among WSIs can aid diagnosis an unknown pathological by finding its similar regions in with However, huge size complex content WSI pose several challenges for retrieval. this paper, we propose unsupervised, accurate, fast method a breast histopathological image. Specifically, presents local statistical feature nuclei morphology distribution nuclei, employs...

Knowledge graph entity typing (KGET) aims to infer missing instances in KGs, which is a significant subtask of KG completion. Despite its progress, however, it still faces two non-trivial challenges: (i) most existing KGET methods extract features by encoding the tuples, while ignoring rich relational knowledge. (ii) they typically treat each tuple KGs independently, and thus inevitably fail take account inherent valuable neighborhood information surrounding tuple. To address these...

Content-based image retrieval (CBIR) has been widely researched for histopathological images. It is challenging to retrieve contently similar regions from whole slide images (WSIs) of interest (ROIs) in different size. In this paper, we propose a novel CBIR framework database that consists WSIs and size-scalable query ROIs. Each WSI the encoded into matrix binary codes. When retrieving, group region proposals have size with ROI are firstly located through an efficient table-lookup approach....

The positive-unlabeled (PU) classification is a common scenario in real-world applications such as healthcare, text classification, and bioinformatics, which we only observe few samples labeled "positive" together with large volume of "unlabeled" that may contain both positive negative samples. Building robust classifiers for the PU problem very challenging, especially complex data where overwhelm mislabeled or corrupted features exist. To address these three issues, propose learning...

Temporal Knowledge Graph (TKG) reasoning involves predicting future facts based on historical information by learning correlations between entities and relations. Recently, many models have been proposed for the TKG task. However, most existing cannot efficiently utilize information, which can be summarized in two aspects: 1) Many only consider a fixed time range, resulting lack of useful information; 2) some use all facts, thus noise or invalid are introduced during reasoning. In this...

Compositional semantic aims at constructing the meaning of phrases or sentences according to compositionality word meanings. In this paper, we propose synchronously learn representations individual words and extracted high-frequency phrases. Representations are considered as gold standard for more general operations compose representation unseen We a grammatical type specific model that improves composition flexibility by adopting vector-tensor-vector operations. Our embodies compositional...

AbstractThe vision of the Semantic Web is to build a global machine-readable data be consumed by intelligent applications. As first step make this come true, initiative linked open has fostered many novel applications aimed at improving accessibility in public Web. Comparably, enterprise environment so different from that most potentially usable business information originates an unstructured form (typically free text), which poses challenge for adoption semantic technologies environment....

Embedding learning on knowledge graphs (KGs) aims to encode all entities and relationships into a continuous vector space, which provides an effective flexible method implement downstream knowledge-driven artificial intelligence (AI) natural language processing (NLP) tasks. Since KG construction usually involves automatic mechanisms with less human supervision, it inevitably brings in plenty of noises KGs. However, most conventional embedding approaches inappropriately assume that facts...

To solve the problem of poor detection effect small-scale pedestrian in image, based on existing deep convolutional neural network(CNN)model. We propose an improved algorithm Faster RCNN. First, image features are extracted by CNN, and areas that may contain pedestrians clustering regional Recommendation Network (RPN), Secondly, a multi-layer feature fusion strategy cascade is proposed. The semantic information network can be enhanced combining high level with those low level. Finally,...

Graph Attention Networks (GATs) have proven a promising model that takes advantage of localized attention mechanism to perform knowledge representation learning (KRL) on graph-structure data, e.g., Knowledge Graphs (KGs).While such approaches entities' local pairwise importance, they lack the capability global importance relative other entities KGs.This causes models miss critical information in tasks where is also significant component for task, as learning.To address issue, we allow proper...