Multimodal sentiment analysis has been an active subfield in natural language processing. This makes multimodal tasks challenging due to the use of different sources for predicting a speaker's sentiment. Previous research focused on extracting single contextual information within modality and trying fusion stages improve prediction accuracy. However, factor that may lead poor model performance is this does not consider variability between modalities. Furthermore, existing methods tend...

Photovoltaic (PV) output power is significantly random and fluctuating due to its sensitivity meteorological factors, making PV forecasting a big challenge. Accurate short-term plays crucial role for the stable operation maintenance management of systems. To achieve this target, paper proposes novel Spatial-Temporal Genetic-based Attention Networks (STGANet), which consists spatial-temporal module (STM) genetic-based attention (GAM). STM serves predict missing solar irradiance support...

Abstract Now deep learn-based object detection can be deployed on drones for criminalinvestigation or military counter-terrorism. Because the proportion of pixels pedestrians orvehicles in aerial picture taken by UAV is very small, probability smallobjects distance low there are omissions. In this paper, HPS-YOLOv7 algorithmis proposed to improve accuracy small objects. We have a modifiedhigh-efficiency layer aggregation network feature extraction, solved problem that theconvergence depth...

With the rapid development of social media, it is very important for automatic rumor detection to obtain global information about propagation. However, get on propagation, necessary use propagation structure rational. In this paper, we propose a novel model, named depth-breadth tree-structured recursive neural networks(RvNN-DB). tree can be used extract depth features multi-branch and breadth at same depth, effectively obtaining propagation; extracting features, introduce attention mechanism...

Multimodal sentiment analysis is an active subfield of natural language processing. It aims to extract and integrate semantic information gathered from multiple modalities identify the sentiments expressed by users. Indeed, complementary heterogeneous between influences prediction results. Recent research proposals employ a single neural network obtain mutually independent representations all modalities. However, problem that may limit previous work reaching higher level this does not...

Most of the existing fake news detection methods focus on feature information at performance level, and pay insufficient attention to semantic content, which cannot fully obtain in news. Therefore, this paper proposes a multi-level enhanced framework (MLSED), imitates way modern people read Aiming two main aspects including entity objects event topics, core content is gradually obtained by mutual enhancement detect Extensive experiments real datasets show that MLSED can capture for...

Multimodal humor detection is an active research area that uses different combinations of modalities for understanding user-generated videos. Previous studies often use a single neural network to process multiple together, which ignores the differences between modalities. Moreover, because there no effective and sufficient information interaction networks, thus complementary not captured. In this paper, we propose new framework, CAMC, first appropriate networks then models multimodal...

Abstract Background: LncRNAs (Long non-coding RNAs) are a type of RNA molecule with transcript length longer than 200 nucleotides. LncRNA has been novel candidate biomarkers in cancer diagnosis and prognosis. However, it is difficult to discover the true association mechanism between lncRNAs complex diseases. The unprecedented enrichment multi-omics data rapid development machine learning technology provide us opportunity design framework study relationship Results: In this article, we...

Flood disasters often occur in summer and autumn. control refers to flood prevention. Corresponding measures means shall be taken according laws characteristics of mitigate or avoid hazards ensure the economic development cities regions. For some similar underground garages, unattended substations other sites, this paper proposes design an automatic gate, which is great significance normal safe operation above sites. The combination ACA (Ant Colony Algorithm) genetic algorithm applied...

Multi-modal entity alignment aims to identify equivalent entities between two individual multi-modal knowledge graphs, and this technique has an essential role in integrating from different data sources. However, most previous works directly adopt simple concatenation or weighted sums as their fusion strategy, ignoring the inter-modal interactions of entities, which leads potential noise introduced uni-modal features during feature stage. The reason for is that use separate encoders encode...

Socialmedia is highly active and full of rumors while the information widely propagation. Mastering process rumor propagation plays an important role in promoting automated detection. However, existing methods ignore evolution development process, resulting loss details, which not con-ducive to extracting key In this paper, we propose a novel hierarchical dynamic graph convolution network (H-DynGCN-Enh) build for each news differentiation, taking into account bi-directional features...

In view of the complex event detection massive data, there are problems such as excessive memory consumption, time stamp sequence and detection.This paper presents an extended hash structure model method to detect detection, It is on basis structure. The events stored according chronological order type, cleaned up period.The efficiency improved while reducing usage.The experimental results show reliability effectiveness method, This makes for some shortcomings existing methods greatly...

In the study, an online prediction algorithm with variable selection and optimization for nonlinear modeling of fast-changing data steams is proposed. The proposed iterative three-step approach. Firstly, extreme learning machine (ELM) trained initial dataset. Secondly, garrote shrinkage operator are introduced to shrink weights input hidden neurons ELM. Thirdly, through update mechanism sequence (OS-ELM), output value ELM after network structural parameters updated iteratively adapt new...

Image-text matching plays a crucial role in connecting vision and language. The details of the objects image, positional relationship, correspondence between background text description are keys to image-text matching. Previous studies either only extract salient or pay attention location object, ignoring detailed features extraction overall semantic information image is not comprehensive enough. Accordingly, this paper proposes model based on Bert Self-Attention Mechanism (BSAM), we segment...

In recent years, public opinion events such as "fake news" and "news reversal" have occurred frequently, spreading rumors through images has become a new form of rumor circulating in the digital age. Most existing methods only consider text content, ignoring role information additional images; for fusion between multiple modalities, their adequate cannot be fully utilized, graphic not interacted. Therefore, we propose multi-level image-text method (MLFRD), which can effectively obtain local...

A radial basis function neural network (RBF) modified maximum correntropy Kalman filter algorithm (RBFMCKF algorithm) is proposed for robot uncalibrated visual servo positioning control. Because there may be non-Gaussian measurement noise in servo, the criterion (MCC) introduced into KF framework to suppress impact of on filtering accuracy, and then RBF used adjusting error produced by (MCKF) algorithm. The simulation results show that RBFMCKF effective can effectively influence non Gaussian...