Quality-of-service (QoS) data vary over time, making it vital to capture the temporal patterns hidden in such dynamic for predicting missing ones with high accuracy. However, currently latent factor (LF) analysis-based QoS-predictors are mostly defined on static QoS without consideration of dynamics. To address this issue, paper presents a biased non-negative factorization tensors (BNLFTs) model pattern-aware prediction. Its main idea is fourfold: 1) incorporating linear biases into...

Visual Semantic Embedding (VSE) is a dominant approach for vision-language retrieval, which aims at learning deep embedding space such that visual data are embedded close to their semantic text labels or descriptions. Recent VSE models use complex methods better contextualize and aggregate multi-modal features into holistic embeddings. However, we discover surprisingly simple (but carefully selected) global pooling functions (e.g., max pooling) outperform those models, across different...

A <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">d</u> ynamically xmlns:xlink="http://www.w3.org/1999/xlink">w</u> eighted irected xmlns:xlink="http://www.w3.org/1999/xlink">n</u> etwork (DWDN) is frequently encountered in various big data-related applications like a terminal interaction pattern analysis system (TIPAS) concerned this study. It consists of large-scale dynamic interactions among numerous nodes. As the involved nodes...

A High-Dimensional and Incomplete (HDI) tensor is frequently encountered in a big data-related application concerning the complex dynamic interactions among numerous entities. Traditional factorization-based models cannot handle an HDI efficiently, while existing latent factorization of tensors are all linear unable to model tensor&#x0027;s nonlinearity. Motivated by this critical discovery, paper proposes Neural Latent Factorization Tensors model, which provides novel approach nonlinear...

Estimating the travel time of a path is great importance to smart urban mobility. Existing approaches are either based on estimating cost each road segment which not able capture many cross-segment complex factors, or designed heuristically in non-learning-based way fail leverage natural abundant temporal labels data, i.e., stamp trajectory point. In this paper, we new development deep neural networks and propose novel auxiliary supervision model, namely DeepTravel, that can automatically...

We propose the Unified Visual-Semantic Embeddings (Unified VSE) for learning a joint space of visual representation and textual semantics. The model unifies embeddings concepts at different levels: objects, attributes, relations, full scenes. view sentential semantics as combination semantic components such objects relations; their are aligned with image regions. A contrastive approach is proposed effective this fine-grained alignment from only image-caption pairs. also present simple yet...

Dynamic relationships are frequently encountered in big data and services computing-related applications, like dynamic of user-side QoS Web services. They modeled into a high-dimensional sparse (HiDS) tensor, which contain rich knowledge regarding temporal patterns. A non-negative latent factorization tensors (NLFT) model is very effective extracting such patterns from an HiDS tensor. However, it commonly suffers overfitting with improper regularization schemes. To address this issue,...

A large-scale dynamically weighted directed network (DWDN) involving numerous entities and massive dynamic interaction is an essential data source in many big-data-related applications, like a terminal pattern analysis system (TIPAS). It can be represented by high-dimensional incomplete (HDI) tensor whose entries are mostly unknown. Yet such HDI contains wealth knowledge regarding various desired patterns potential links DWDN. latent factorization-of-tensors (LFT) model proves to highly...

We consider the problem of learning distributed representations for documents in data streams. The are represented as low-dimensional vectors and jointly learned with vector word tokens using a hierarchical framework two embedded neural language models. In particular, we exploit context streams use one models to model document sequences, other sequences within them. learn continuous both such that semantically similar words close common space. discuss extensions our model, which can be...

Trajectory outlier detection is a fundamental building block for many location-based service (LBS) applications, with large application base. We dedicate this paper on detecting the outliers from vehicle trajectories efficiently and effectively. In addition, we want our solution to be able issue an alarm early when trajectory only partially observed (i.e., has not yet reached destination). Most existing works study problem general Euclidean require accesses historical database or...

A nonnegative latent factorization of tensors (NLFT) model precisely represents the temporal patterns hidden in multichannel data emerging from various applications. It often adopts a single factor-dependent, and multiplicative update on tensor (SLF-NMUT) algorithm. However, learning depth this algorithm is not adjustable, resulting frequent training fluctuation or poor convergence caused by overshooting. To address issue, study carefully investigates connections between performance an NLFT...

Vehicle trajectories are one of the most important data in location-based services. The quality directly affects However, real applications, trajectory not always sampled densely. In this paper, we study problem recovering entire route between two distant consecutive locations a trajectory. Most existing works solve without using those informative historical or it an empirical way. We claim that data-driven and probabilistic approach is actually more suitable as long sparsity can be well...

Abstract This article studies the secure leader‐following consensus control problem for discrete‐time multi‐agent systems under false data injection (FDI) attacks on both actuators and sensors. It is shown that standard protocol, an attacked follower can cause normal followers to fail follow leader if access through directed paths of arbitrary length. To mitigate adverse impacts FDI attacks, a observer‐based protocol presented where local observer used estimate system state attack signal...

Text-supervised semantic segmentation is a novel research topic that allows segments to emerge with image-text contrasting. However, pioneering methods could be subject specifically designed network architectures. This paper shows vanilla contrastive language-image pretraining (CLIP) model an effective text-supervised segmentor by itself. First, we reveal CLIP inferior localization and due its optimization being driven densely aligning visual language representations. Second, propose the...

Abstract Background Ustekinumab (UST), a novel biological agent for Crohn’s disease (CD), has shown efficacy in CD patients. Our objective was to establish clinical prediction models accurately forecast the short- and long-term effectiveness of UST individuals diagnosed with CD. Methods We established derivation cohort comprising patients treated at Sixth Affiliated Hospital Sun Yat-sen University from May 2020 July 2023. The were assigned 7:3 ratio random training validation cohort....

Abstract Background Endoscopic balloon dilation (EBD) is a safe and effective procedure for treating stenosis in patients with Crohn's disease (CD). This study aimed to evaluate factors associated endoscopic restenosis after EBD construct prognostic model. Methods We retrospectively collected analyzed data on receiving treatment at the Sixth Affiliated Hospital of Sun Yat-sen University from 2013 2024. Seven machine learning (ML) algorithms were used And explore potential biomarkers...

Abstract Background Ustekinumab (UST) is recommended as the first-line treatment for patients with moderate to severe Crohn's disease (CD). The effectiveness of certain patients, however, may be suboptimal and necessitate intensive or modification regimen. We sought establish a nomogram model predict short-term UST in CD patients. Methods established derivation cohort comprising diagnosed treated at Sixth Affiliated Hospital Sun Yat-sen University between May 2020 July 2023. patient data,...

Abstract Background Inflammatory bowel disease (IBD) requires effective treatment options. Upadacitinib, a Janus kinase 1 (JAK1) inhibitor, has shown effectiveness in trials for Crohn’s (CD) and ulcerative colitis (UC). This study evaluates its real-world safety. Methods We conducted multicenter retrospective cohort tertiary care centers, involving patients treated with upadacitinib from January 2023 to September 2024. The included adult aged 18 years or older, diagnosed UC CD, who received...

Spatio-temporal trajectory classification is a fundamental problem for location-based services with many real-world applications such as travel mode classification, animal mobility detection, and location recommendation. In the literature, approaches have been proposed to solve this task including deep learning models like LSTM recently sequence classification. However, these fail consider both spatial temporal interval information simultaneously, but share some common drawbacks: omitting...

The medical crowdsourced question answering (Q&A) websites are booming in recent years, and an increasingly large amount of patients doctors involved. valuable information from these Q&A can benefit patients, the society. One key to unleash power is extract knowledge noisy question-answer pairs filter out unrelated or even incorrect information. Facing daunting scale generated on everyday, it unrealistic fulfill this task via supervised method due expensive annotation cost. In paper, we...