E-commerce users may expect different products even for the same query, due to their diverse personal preferences. It is well known that there are two types of preferences: long-term ones and short-term ones. The former refers users’ inherent purchasing bias evolves slowly. By contrast, latter reflects inclination in a relatively short period. They both affect current intentions. However, few research efforts have been dedicated jointly model them personalized product search. To this end, we...

Recent years have witnessed a flourishing of community-driven question answering (cQA), like Yahoo! Answers and AnswerBag, where people can seek precise information. After 2010, some novel cQA systems, including Quora Zhihu, gained momentum. Besides interactions, the latter enables users to label questions with topic tags that highlight key points conveyed in questions. In this article, we shed light on automatically annotating newly posted are predefined preorganized into directed acyclic...

Online micro-video recommender systems aim to address the information explosion of micro-videos and make personalized recommendation for users. However, existing methods still have some limitations in learning representative user interests, since multi-scale time effects, interest group modeling, false positive interactions are not taken into consideration. In view this, we propose an end-to-end Multi-scale Time-aware Interest modeling Network (MTIN). particular, first present routing...

Multicasting does not work well for video applications such as on demand, where individual users request content at discrete and unpredictable moments. In this paper, we study a concurrent multicast orchestration solution the caching-assisted centrally controlled mobile networks. The proposed strives to obtain rapid prefetch effective traffic reduction. By prefetch, user devices can cache required contents in advance, which effectively improve streaming fluency unsteady wireless...

Conversational image search, a revolutionary search mode, is able to interactively induce the user response clarify their intents step by step. Several efforts have been dedicated conversation part, namely automatically asking right question at time for preference elicitation, while few studies focus on part given well-prepared conversational query. In this paper, we work towards which much difficult compared traditional task, due following challenges: 1) understanding complex from...

Video moment retrieval, i.e., localizing the specific video moments within a given description query, has attracted substantial attention over past several years. Although great progress been achieved thus far, most of existing methods are supervised, which require moment-level temporal annotation information. In contrast, weakly-supervised only need video-level annotations remain largely unexplored. this paper, we propose novel end-to-end Siamese alignment network for retrieval. To be...

By leveraging deep neural networks, recent face swapping techniques have performed admirably in generating faces that maintain consistent identities. Nevertheless, while these methods accurately transfer source identities, they often struggle to preserve important attributes (such as head poses, expressions, and gaze directions) the target faces. As a consequence, current research this domain has not resulted satisfactory performance. In paper, we propose an efficient attribute-preserving...

Recently, as an essential part of people's daily life, clothing matching has gained increasing research attention. Most existing efforts focus on the numerical compatibility modeling between fashion items with advanced neural networks, and hence suffer from poor interpretation, which makes them less applicable in real world applications. In fact, people prefer to know not only whether given are compatible, but also reasonable interpretations well suggestions regarding how make incompatible...

Combining medical data and machine learning has fully utilized the value of data. However, contain a large amount sensitive information, inappropriate handling can lead to leakage personal privacy. Thus, both publishing training in may reveal privacy patients. To address above issue, we propose two effective approaches. One combines differential decision tree (DPDT) approach provide strong guarantees for data, which establishes weight calculation system based on classification regression...

Video moment localization, as an important branch of video content analysis, has attracted extensive attention in recent years. However, it is still its infancy due to the following challenges: cross-modal semantic alignment and localization efficiency. To address these impediments, we present a network. be specific, first design encoder generate candidates, learn their representations, well model relevance. Meanwhile, query for diverse intention understanding. Thereafter, introduce...

A deep learning-based Visual Inertial SLAM technique is proposed in this paper to ensure accurate autonomous localization of mobile robots environments with dynamic objects. Addressing the limitations real-time performance learning algorithms and poor robustness pure visual geometry algorithms, presents a technique. Firstly, non-blocking model designed extract semantic information from images. Then, motion probability hierarchy obtain prior probabilities feature points. For image frames...

Nodes in wireless sensor networks (WSNs) are usually deployed an unattended even hostile environment. What is worse, these nodes equipped with limited battery, storage, computation, and communication resources. Therefore, it challenging to ensure the security of a WSN without decreasing its network performance. Data aggregation (DA) combined mechanism can provide good scheme for solving aforementioned problems. This article presents comprehensive review secure DA (SDA) WSNs, including goals...

Benefiting from the advancement of computer vision, natural language processing and information retrieval techniques, visual question answering (VQA), which aims to answer questions about an image or a video, has received lots attentions over past few years. Although some progress been achieved so far, several studies have pointed out that current VQA models are heavily affected by prior problem, means they tend based on co-occurrence patterns keywords (e.g., how many) answers 2) instead...

Atrial fibrillation (AF) is one of the most common persistent arrhythmias, which has a close connection to large number cardiovascular diseases. However, if spotted early, diagnosis AF can improve effectiveness clinical treatment and effectively prevent serious complications. In this paper, combination an 8-layer convolutional neural network (CNN) with shortcut 1-layer long short-term memory (LSTM), named 8CSL, was proposed for Electrocardiogram (ECG) classification task. Compared recurrent...

The 12-lead electrocardiogram (ECG) is widely used in the clinical diagnosis of cardiovascular disease, and deep learning has become an effective approach to automatic ECG classification. Generally, current research simply regards signal as ordinary 2-D array does not specifically consider intrinsic relationship between different leads when building neural networks. However, classes, from biomedical perspective, mainly show specific patterns on one or several rather than all 12 leads, which...

To reduce the influence of both baseline wander (BW) and noise in electrocardiogram (ECG) is much important for further analysis diagnosis heart disease. This paper presents a convex optimization method, which combines linear time-invariant filtering with sparsity BW correction denoising ECG signals. The signals are modeled as low-pass signals, while sequence sparse have derivatives. illustrate positive peaks, an asymmetric function symmetric used to punish original their difference...

Multimodal dialog system has attracted increasing attention from both academia and industry over recent years. Although existing methods have achieved some progress, they are still confronted with challenges in the aspect of question understanding (i.e., user intention comprehension). In this paper, we present a relational graph-based context-aware scheme, which enhances comprehension local to global. Specifically, first utilize multiple attribute matrices as guidance information fully...

High-quality and high-fidelity removal of noise in the Electrocardiogram (ECG) signal is great significance to auxiliary diagnosis ECG diseases. In view single function traditional denoising methods insufficient performance details after denoising, a new method based on combination Generative Adversarial Network (GAN) Residual proposed. The adopted this paper GAN structure, it restructures generator discriminator. network, residual blocks Skip-Connecting are used deepen network structure...

A large number of resources are integrated into a data center to provide various resource services in cloud computing. major challenge is how timely and accurately satisfy users’ demands. However, demands change constantly sometimes fluctuate very strong. The provision may be not performed time. And even, the active physical too insufficient because some them shut down order reduce energy. So it important proactive guarantee good experiences key predict future support advance. In this paper,...

Dynamic resource scheduling is a critical activity to guarantee quality of service (QoS) in cloud computing. One challenging problem how predict future host utilization real time. By predicting utilization, data center can place virtual machines suitable hosts or migrate advance from overloaded underloaded QoS save energy. However, it very difficult accurately timely manner because varies quickly and exhibits strong instability with many bursts. Although machine learning methods usually...

Abstract With the development of big data and artificial intelligence, cloud resource requests present more complex features, such as being sudden, arriving in batches diverse, which cause allocation to lag far behind an unbalanced utilization that wastes resources. To solve this issue, paper proposes a proactive method based on adaptive prediction computing. Specifically, first runs test improves accuracy requests, then, it builds multiobjective optimization model, alleviates latency...