Multi-view semi-supervised classification (Mv-SSC) aims to leverage complementary information from multiple views and unlabeled data to enhance classification performance. However, existing methods often struggle with constructing robust graphs, effectively fusing heterogeneous features. Therefore, we propose a novel framework, Multi-view Semi-supervised Classification with Graph Construction Innovation and Smoothness-aware Graph Convolution (GCSGC), which introduces three key innovations. First, we employ sparse autoencoders (SAEs) coupled with a learnable feature fusion module based on a one-layer fully connected network. The SAEs extract compact and discriminative representations from each view by enforcing sparsity to remove redundancies, while the fully connected layer adaptively aligns the multi-view latent features with the learnable shared features to preserve complementary semantic information. Second, GCSGC introduces two types of view-based graphs: (1) a Cosine-KNN-based Collaborative Graph (CKG) that integrates K-Nearest Neighbor relationships with cosine similarity of the learned view-specific latent features, and (2) a robust semi-supervised graph that enhances inter-view consistency and structural stability. Third, we design a hybrid loss function that combines cross-entropy and smoothness regularization to guide the optimization of the graph convolutional network. Extensive experiments on multiple benchmark datasets demonstrate that GCSGC consistently outperforms state-of-the-art Mv-SSC methods, validating the effectiveness of its innovative graph construction strategy, sparse feature fusion mechanism, and dual-loss optimization. This work provides a new perspective on integrating structural modeling, feature transformation, and loss design in multi-view semi-supervised learning.

The networking industry is offering new services leveraging recent technological advances in connectivity, storage, and computing such as mobile communications and edge computing. In this regard, extended reality, a term encompassing virtual reality, augmented reality, and mixed reality, can provide unprecedented user experience and pioneering service opportunities such as: live concerts, sports, and other events; interactive gaming and entertainment; immersive education, training, and demos. These services require high-bandwidth, low-latency, and reliable connections, and are supported by next-generation ultra-reliable and low-latency communications in the vision of 6G mobile communication systems. In this work, we devise a novel scheme, called backup from different data centers with multicast and adaptive bandwidth provisioning, to admit reliable, low-latency, and high-bandwidth extended reality live streams in next-generation networks. We consider network services where contents are non-cacheable and investigate how backup services can be offered by different data centers with multicast and adaptive bandwidth provisioning. Our proposed service-provisioning scheme provides protection not only against link failures in the physical network but also against computing and storage failures in data centers. We develop scalable algorithms for the service-provisioning scheme and evaluate their performance on various complex network instances in a dynamic environment. Numerical results show that, compared to conventional service-provisioning schemes such as those seeking backup services from the same data center, our proposed service-provisioning scheme efficiently utilizes network resources, ensures higher reliability, and guarantees low latency; hence, it is highly suitable for extended reality live streams.

Data augmentation can mitigate overfitting problems in data exploration without increasing the size of the model. Existing cutmix-based data augmentation has been proven to significantly enhance deep learning performance. However, many existing methods overlook the discriminative local context of the image and rely on ad hoc regions consisting of square or rectangular local regions, resulting in the loss of complete semantic object parts. In this work, we propose a superpixelwise local-context-aware efficient image mixing approach for data augmentation, aiming to overcome the limitations previously mentioned. Our approach only requires one forward propagation using a superpixel attention-based label mixing with lower computational complexity. The model is trained using a combination of a global classification of the mixed (augmented) image loss, a superpixel-wise weighted local classification loss, and a superpixel-based weighted contrastive learning loss. The last two losses are based on the superpixel-aware attentive embeddings. Thus, the resulting deep encoder can learn both local and global features of the images, capturing object-part local context information. Experiments on diverse benchmarks, such as ImageNet-1K and CUB-200-2011, indicate that the proposed method out-performs many augmentation methods for visual recognition. We have not only demonstrated its effectiveness on CNN models, but also on transformer models.

The exponential growth of the Internet of Multimedia Things (IoMT) traffic has posed a threat of service quality degradation due to the limitation of current communication, networking, and computing advances in mobile networks. In this regard, managing the Quality-of-Experience (QoE) for IoMT services is a vital challenge to meet user satisfaction. To cope with this problem, we investigate the joint optimization of video quality variation and latency in multiuser downlink rate-splitting multiple-access (RSMA) networks, especially within imperfect network conditions and state information. To accomplish this, we first formulated the joint optimization problem into a Markov decision process framework, then exploited a deep reinforcement learning approach to adaptively calculate the optimal configuration of the RSMA against environment dynamics. As a result, the proposed deep deterministic policy gradient on RSMA-based video streaming system (DDPG-RMAVS) provides QoE maintenance by minimizing video resolution reduction and latency. Extensive simulation results revealed that the proposed DDPG-RMAVS algorithm surpasses existing algorithms by achieving higher video quality, lower delay, larger buffer capacity, and limited stalling events, representing a significant breakthrough in IoMT streaming optimization.

Kinship verification is an interesting topic of face analysis. In this paper, we propose to extract several local image descriptors including Local Binary Pattern, Local Ternary Pattern and Histogram of Gradient for extracting features from facial images. Next, we apply an extension of constraint score for features ranking and then select a subset of feature for training by Support Vector Machine. The fusion at an early stage (feature level) and fusion at a later stage (score level) are compared and evaluated on two benchmark kinship datasets KinFaceW-I and KinFaceW-II. Experimental results show that the proposed approach outperforms other works in the state-of-the-art.

Analysis of lace texture images is a challenging problem because the lace is a soft and extensible material and can be easily deformed. This paper investigates a whole system for lace classification. A first step, based on Otsu's segmentation method, allows to remove the background. Then the lace texture is characterized using local binary patterns (LBP). In order to be robust against rotation the Fourier Transform is applied on LBP histograms. The magnitude spectrum of this transform is then used as a feature vector. LBP descriptor parameters, including radius and number of neighbors, are adjusted in order to improve their relevance. The experiments show that the features based on LBP, with appropriate settings, produced good results in supervised and unsupervised contexts.