Seeking multiple optima simultaneously, which multimodal optimization aims at, has attracted increasing attention but remains challenging. Taking advantage of ant colony (ACO) algorithms in preserving high diversity, this paper intends to extend ACO deal with optimization. First, combined current niching methods, an adaptive continuous algorithm is introduced. In algorithm, parameter adjustment developed, takes the difference among niches into consideration. Second, accelerate convergence, a...

In a heterogeneous Wireless Sensor Network (WSN), factors such as initial energy, data processing capability, etc. greatly influence the network lifespan. Despite success of various clustering strategies WSN, numerous possible sensor clusters make searching for an optimal structure open challenge. this paper, we propose Genetic Algorithm based method that optimizes node clustering. Compared with five state-of-the-art methods, our proposed extends life, and average improvement respect to...

Research on social bot detection plays a crucial role in maintaining the order and reliability of information dissemination while increasing trust interactions. The current mainstream models rely black-box neural network technology, e.g., Graph Neural Network, Transformer, etc., which lacks interpretability. In this work, we present UnDBot, novel unsupervised, interpretable, yet effective practical framework for detecting bots. This is built upon structural theory. We begin by designing...

In this study, we propose Multimodal Fusion-supervised Cross-modality Alignment Perception (MulFS-CAP), a novel framework for single-stage fusion of unregistered infrared-visible images. Traditional two-stage methods depend on explicit registration algorithms to align source images spatially, often adding complexity. contrast, MulFS-CAP seamlessly blends implicit with fusion, simplifying the process and enhancing suitability practical applications. utilizes shared shallow feature encoder...

Existing differential evolution (DE) algorithms often face two challenges. The first is that the optimization performance significantly affected by ad hoc configurations of operators and parameters for different problems. second long runtime real-world problems whose fitness evaluations are expensive. Aiming at solving these problems, this paper develops a novel double-layered heterogeneous DE algorithm realizes it in cloud computing distributed environment. In layer, populations with...

Person re-identification aims to match the images of pedestrians across different camera views from locations. This is a challenging intelligent video surveillance problem that remains an active area research due need for performance improvement. involves two main steps: feature representation and metric learning. Although keep it simple straightforward (KISS) learning method discriminative distance has been shown be effective person re-identification, estimation inverse covariance matrix...

This paper presents a novel opinion mining research problem, which is called Contrastive Opinion Modeling (COM). Given any query topic and set of text collections from multiple perspectives, the task COM to present opinions individual perspectives on topic, furthermore quantify their difference. general problem subsumes many interesting applications, including summarization forecasting, government intelligence cross-cultural studies. We propose unsupervised model for contrastive modeling. It...

Person re-identification aims to identify the same pedestrians across different camera views at locations. This important yet difficult intelligent video analysis problem remains a vigorous area of research due demands for performance improvements. involves two main steps: feature representation and metric learning. Handcrafted features, such as color texture histograms, are frequently used person re-identification, but most handcrafted features limited by not being directly applicable...

Crowdsourcing labeling systems provide an efficient way to generate multiple inaccurate labels for given observations. If the competence level or "reputation," which can be explained as probabilities of annotating right label, each crowdsourcing annotators is equal and biased annotate majority voting (MV) optimal decision rule merging into a single reliable one. However, in practice, levels employed by are often diverse very much. In these cases, weighted MV more preferred. The weights...

Domain invariance and discrimination of learned features as two crucial factors affect the performance unsupervised domain adaptation (UDA) person re-identification (Re-ID). Person attributes (such "backpack", "boots", "handbag", etc) remaining unchanged across multiple domains have been used mid-level visual-semantic information in UDA Re-ID. As main challenges, both misalignment attribute-related regions images shift between source target learning domain-invariant (DIF). To address above...

Due to the importance of practical applications, unsupervised domain adaptation (UDA) person re-identification (re-ID) has attracted increasing attention. However, most existing methods often lack multi-view information reasoning and ignore discrepancy pedestrian images with same identity, which constrain further improvement recognition performance. So, this paper proposes a triple adversarial learning imaginative network (TAL-MIRN) for UDA re-ID, consists module (IRM) (TALM). IRM makes...

This article presents a novel extended reality (XR) and deep-learning-based Internet-of-Medical-Things (IoMT) solution for the COVID-19 telemedicine diagnostic, which systematically combines virtual reality/augmented (AR) remote surgical plan/rehearse hardware, customized 5G cloud computing deep learning algorithms to provide real-time treatment scheme clues. Compared existing perception therapy techniques, our new technique can significantly improve performance security. The system...

As a trending approach for social event detection, graph neural network (GNN)-based methods enable fusion of natural language semantics and the complex structural information, thus showing SOTA performance. However, GNN-based can miss useful message correlations. Moreover, they require manual labeling training predetermining number events prediction. In this work, we address detection via entropy (SE) minimization. While keeping merits methods, proposed framework, HISEvent, constructs more...

Current infrared and visible image fusion (IVIF) methods go to great lengths excavate complementary features design complex strategies, which is extremely challenging. To this end, we rethink the IVIF outside box, proposing a harmonious information transfer network (CHITNet). It reasonably transfers into one, integrates both shared from two modalities. Specifically, skillfully sidestep aggregating in IVIF, mutual (MIT) module mutually represent modalities, roughly transferring one. Then,...

Community detection is a critical task in graph theory, social network analysis, and bioinformatics, where communities are defined as clusters of densely interconnected nodes. However, detecting large-scale networks with millions nodes billions edges remains challenging due to the inefficiency unreliability existing methods. Moreover, many current approaches limited specific types, such unweighted or undirected graphs, reducing their broader applicability. To address these issues, we propose...

Due to the domain shift between source dataset and target dataset, most of existing person re-identification (PRID) algorithms trained by a supervised learning framework often fail be well generalized another domain. To address this challenge, we propose self-supervised algorithm based on attribute-identity embedding, which can incrementally optimize model selecting unlabeled samples from Thus gap is bridged. Specifically, first develop an joint prediction dictionary for simultaneously...

Due to incomplete appearance features, the identity matching of occluded pedestrians under multiple cross-camera views is a long-term challenge. Although existing re-identification (re-ID) solutions have made significant progress, most them achieve accurate by extracting pedestrian features from unoccluded areas. However, when partially blocked body another pedestrian, methods cannot accurately determine whether parts belong target which brings great difficulties matching. To alleviate this...