Near-duplicate video retrieval (NDVR) has been a significant research task in multimedia given its high impact applications, such as search, recommendation, and copyright protection. In addition to accurate performance, the exponential growth of online videos imposed heavy demands on efficiency scalability existing systems. Aiming at improving both accuracy speed, we propose novel stochastic multiview hashing algorithm facilitate construction large-scale NDVR system. Reliable mapping...

Unsupervised Cross-domain Sentiment Classification is the task of adapting a sentiment classifier trained on particular domain ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">source domain</i> ), to different xmlns:xlink="http://www.w3.org/1999/xlink">target without requiring any labeled data for target domain. By an existing previously unseen domains, we can avoid cost manual annotation We model this problem as embedding learning, and...

Everyone's walking style is unique, and it has been shown that both humans computers are very good at recognizing known gait patterns. It therefore unsurprising dynamic foot pressure patterns, which indirectly reflect the accelerations of all body parts, also previous studies have achieved moderate-to-high classification rates (CRs) using variables. However, these limited by small sample sizes (n < 30), moderate CRs (CR ≃ 90%), or both. Here we show, relatively simple image processing...

Route planning for fully electric vehicles (FEVs) must take energy efficiency into account due to limited battery capacity and time-consuming recharging. In addition, the algorithm should allow negative costs in road network regenerative braking, which is a unique feature of FEVs. this paper, we propose framework energy-driven context-aware route It has two novel aspects: 1) context aware, i.e., access real-time traffic data routing cost estimation; it driven, both time are accounted for;...

Zero-shot learning (ZSL) suffers intensely from the domain shift issue, i.e., mismatch (or misalignment) between true and learned data distributions for classes without training (unseen classes). By additionally unlabelled collected unseen classes, transductive ZSL (TZSL) could reduce but only to a certain extent. To improve TZSL, we propose novel approach Bi-VAEGAN which strengthens distribution alignment visual space an auxiliary space. As result, it can largely shift. The proposed key...

We propose two variations of the support vector data description (SVDD) with negative samples (NSVDD) that learn a closed spherically shaped boundary around set in target class by involving different forms slack vectors, including two-norm NSVDD and nu-NSVDD. extend NSVDDs to solve multiclass classification problems based on distances between centers learned boundaries kernel-defined feature space using combination linear discriminant analysis (LDA) nearest-neighbor (NN) rule. Extensive...

In this paper, a novel unsupervised hashing algorithm, referred to as t-USMVH, and its extension deep hashing, t-UDH, are proposed support large-scale video-to-video retrieval. To improve robustness of the learning, t-USMVH combines multiple types feature representations effectively fuses them by examining continuous relevance score based on Gaussian estimation over pairwise distances, also discrete neighbor cardinality reciprocal neighbors. reduce sensitivity scale changes for mapping...

Sentiment analysis is an important topic concerning identification of feelings, attitudes, emotions and opinions from text. To automate such analysis, a large amount example text needs to be manually annotated for model training. This laborious expensive, but the cross-domain technique key solution reducing cost by reusing reviews across domains. However, its success largely relies on learning robust common representation space In recent years, significant effort has been invested improve...

We present a theory of ensemble diversity, explaining the nature diversity for wide range supervised learning scenarios. This challenge has been referred to as holy grail learning, an open research issue over 30 years. Our framework reveals that is in fact hidden dimension bias-variance decomposition loss. prove family exact bias-variance-diversity decompositions, losses both regression and classification, e.g., squared, cross-entropy, Poisson losses. For where additive not available (e.g.,...

Citation screening, an integral process within systematic reviews that identifies citations relevant to the underlying research question, is a time-consuming and resource-intensive task. During screening task, analysts manually assign label each citation, designate whether citation eligible for inclusion in review. Recently, several studies have explored use of active learning text classification reduce human workload involved However, existing approaches require significant amount labelled...

Group-agent reinforcement learning (GARL) is a newly arising scenario, where multiple agents study together in group, sharing knowledge an asynchronous fashion. The goal to improve the performance of each individual agent. Under more general heterogeneous setting different learn using algorithms, we advance GARL by designing novel and effective group-learning mechanisms. They guide on whether how from action choices others, allow adopt available policy value function models sent another...

Compared to conventional semantic segmentation with pixel-level supervision, weakly supervised (WSSS) image-level labels poses the challenge that it commonly focuses on most discriminative regions, resulting in a disparity between and fully supervision scenarios. A typical manifestation is diminished precision object boundaries, leading deteriorated accuracy of WSSS. To alleviate this issue, we propose adaptively partition image content into certain regions (e.g., confident foreground...

The question answering system in open domain enables a machine to automatically select and generate the answer for questions posed by humans natural language form on website. Previous approaches seek effective ways of extracting semantic features between answer, but contextual information effects matching are still limited short-term memory. As an alternative, we propose internal knowledge-based end-to-end model, enhanced attentive memory network both selection generation tasks considering...

Clinical trials are mandatory protocols describing medical research on humans and among the most valuable sources of practice evidence. Searching for relevant to some query is laborious due immense number existing protocols. Apart from search, writing new includes composing detailed eligibility criteria, which might be time-consuming, especially researchers. In this paper we present ASCOT, an efficient search application customised clinical trials. ASCOT uses text mining data methods enrich...

This paper is about supervised and semi-supervised dimensionality reduction (DR) by generating spectral embeddings from multi-output data based on the pairwise proximity information. Two flexible generic frameworks are proposed to achieve DR (SDR) for multilabel classification. One able extend any existing single-label SDR via sample duplication, referred as MESD. The other a design framework that tackles problem computing weight (proximity) matrices simultaneous feature label information,...

The strategy of ensemble has become popular in adversarial defense, which trains multiple base classifiers to defend against attacks a cooperative manner. Despite the empirical success, theoretical explanations on why an adversarially trained is more robust than single ones remain unclear. To fill this gap, we develop new error theory dedicated understanding demonstrating provable 0-1 loss reduction challenging sample sets defense scenario. Guided by theory, propose effective approach...