This article is inspired by video classification technology. If the user behavior subspace is viewed as a frame image, consecutive frame images are viewed as a video. Following this novel idea, a model for spammer detection based on user videoization, called UVSD, is proposed. Firstly, a user2piexl algorithm for user pixelization is proposed. Considering the adversarial behavior of user stances, the user is viewed as a pixel, and the stance is quantified as the pixel’s RGB. Secondly, a behavior2image algorithm is proposed for transforming user behavior subspace into frame images. Low-rank dense vectorization of subspace user relations is performed using representation learning, while cutting and diffusion algorithms are introduced to complete the frame imageization. Finally, user behavior videos are constructed based on temporal features. Subsequently, a video classification algorithm is combined to identify the spammers. Experiments using publicly available datasets, i.e., WEIBO and TWITTER, show an advantage of the UVSD model over state-of-the-art methods.

Multimodal emotion recognition is crucial for future human-computer interaction. However, accurate emotion recognition still faces significant challenges due to differences between different modalities and the difficulty of characterizing unimodal emotional information. To solve these problems, a hybrid network model based on multipath cross-modal interaction (MCIHN) is proposed. First, adversarial autoencoders (AAE) are constructed separately for each modality. The AAE learns discriminative emotion features and reconstructs the features through a decoder to obtain more discriminative information about the emotion classes. Then, the latent codes from the AAE of different modalities are fed into a predefined Cross-modal Gate Mechanism model (CGMM) to reduce the discrepancy between modalities, establish the emotional relationship between interacting modalities, and generate the interaction features between different modalities. Multimodal fusion using the Feature Fusion module (FFM) for better emotion recognition. Experiments were conducted on publicly available SIMS and MOSI datasets, demonstrating that MCIHN achieves superior performance.

Vertical federated learning can aggregate participant data features. To address the issue of insufficient overlapping data in vertical federated learning, this study presents a generative adversarial network model that allows distributed data augmentation. First, this study proposes a distributed generative adversarial network FeCGAN for multiple participants with insufficient overlapping data, considering the fact that the generative adversarial network can generate simulation samples. This network is suitable for multiple data sources and can augment participants' local data. Second, to address the problem of learning divergence caused by different local distributions of multiple data sources, this study proposes the aggregation algorithm FedKL. It aggregates the feedback of the local discriminator to interact with the generator and learns the local data distribution more accurately. Finally, given the problem of data waste caused by the unavailability of nonoverlapping data, this study proposes a data augmentation method called VFeDA. It uses FeCGAN to generate pseudo features and expands more overlapping data, thereby improving the data use. Experiments showed that the proposed model is suitable for multiple data sources and can generate high-quality data.

This study was inspired by video forgery detection techniques. If the topic space at a certain time is considered as a frame image, the consecutive frame images over time could be viewed as a video. Then the rumor topic detection problem is transformed into a topic video forgery detection problem. Thus, a novel rumor detection method was proposed. First, a Topic2RGB algorithm was proposed to convert comment users into pixel points. The algorithm views commenting users as pixel points while using game theory to mine user pro-opposition emotions as RGB information. Secondly, a Topic2Video algorithm was proposed to convert the topic space into video. The algorithm converts the topic space into frame images. Meanwhile, the topic space is time-sliced, then the topic space is transformed into a video. Finally, the volatility of user emotional confrontation during a long time in the topic space is like the change of characteristics of frame images in forgeries videos. Then, a topic video rumor detection method (TVRD) was proposed. The experiments indicate that the method successfully verifies the viability of the topic videolization for rumor detection. Additionally, the method also demonstrates the effectiveness of user emotion confrontation of topic space on detection performance.

Rumors in different topic domains have different text characteristics but similar emotional tendencies. To resolve the scarce-data problem in some rumor-topic domains, this study proposes a cross-domain rumor-propagation model, which is based on transfer learning. First, given the diversity and complexity of the rumor-propagation landscape, this study introduces a novel method, User-Retweet-Rumor2vec (URR2vec), which leverages the power of representation learning to uncover latent features within rumor topics. It also displays the forwarding relationship between users and rumors, user node information, and rumor-topic information in low-dimensional space. To capture the impact of human emotional cognition during rumor spreading, we also introduce a deep-learning model based on the natural language texts of rumor topics, which analyzes the sentiment in the text and uncovers the emotional correlations among users. Furthermore, a rumor-propagation prediction model based on the text-sentiment analysis-graph convolutional network (TSA-GCN) is proposed and pre-trained on existing rumor-topic data to ensure its prediction accuracy. Finally, considering the data sparsity at a rumor-topic outbreak, the trained propagation model is transferred to the rumor topic for prediction. Meanwhile, the rumor topic in different domains has different edges and conditional distribution, similar emotional characteristics, and network structure among the rumor topics. After fine-tuning the parameter and adding a domain adaptation layer in TSA-GCN, a domain adaptation model based on parameter and graph-structure migration is obtained.

This article is inspired by image forgery recognition techniques. If we regard topic comments as image pixels, the whole topic is a complete image. The image differences between rumor topics and nonrumor topics are reflected in image pixels just like forged images, and then, the problem of detecting rumor topics can be regarded as the problem of recognition images of rumor topics. First, the Topic2Image algorithm is proposed to use the semantic information to quantify the adversarial intensity among comments. It is mapped to the topological relationship among user comments. Also, the relative positions of the comment nodes are determined by the adversarial intensity. Second, considering the competitive relationship between positive and negative comments, a sentimental mutual influence model is proposed. Based on the evolutionary game theory, a transfer matrix of sentimental mutual influence is constructed. Internal and external factors of rumor detection are considered at the individual and group levels, respectively. Finally, considering the advantages of convolutional neural network (CNN) for image processing, a simple rumor detection algorithm topic image rumor detection (TIRD) based on topic image classification is proposed. Using CNNs and gray-level co-occurrence matrix to extract global and local features of topic images and combining them with the transfer matrix of sentimental mutual influence, the detection of topic rumor is realized. Experiments demonstrate the feasibility of transforming topic rumors into image. In addition, the effectiveness of image forgery recognition technology for detecting rumors is verified.

Abstract Only a subset of subcarriers are activated in orthogonal frequency division multiplexing‐index modulation (OFDM‐IM), which achieves higher energy efficiency and resists frequency offset. In the OFDM‐IM, the energy of the received signal is computed and then combined with pre‐processed signal to create the input of detection network. Inspired by image denoising technology, this study enhances the detection performance by denoising the pre‐processed data and improving the energy distribution in the OFDM‐IM system. First, considering that the noise reduction process of the pre‐processed signal can effectively mitigate the distortion by noise which affects the detection accuracy, this study proposes a two‐phase neural network termed as Deep‐Denoising‐IM through the combination of a noise reduction network and a deep learning detection method. Then, to better determine the position of the active carriers, a joint decision method of the denoised data and the received signal is designed as the IQ signal of the denoised data may change the quadrant of original signal distribution. In addition, the pre‐processed data sample has insufficient diversity. Considering that data enhancement can increase the noise of the signal samples, this study proposes a method to strengthen the silent carriers in the model training phase, which improves the generalization ability of the model and further enhances the denoising performance. Simulation results show that Deep‐Denoising‐IM outperforms the existing detectors in terms of mean square error (MSE) and bit error rate (BER) under the Rayleigh fading channel.