Virtual ophthalmic surgical training is a cost-effective and time-efficient paradigm. However, the absence of haptic feedback in virtual ophthalmic surgery systems limits the realism and effectiveness of training. To address this challenge, this study proposes a task-driven method tailored to quantify haptic information in ophthalmic cataract surgery. This work designs a virtual cataract surgery interaction scenario and simulates three key haptic interactions: damping when moving the surgical tool within the eyeball, stiffness during corneal incision and phacoemulsification, and pull during continuous curvilinear capsulorhexis. Using a task-driven and subjective-objective integrated data analysis approach, this work quantifies the just noticeable threshold (JNT) and just noticeable difference (JND). Experimental results demonstrate that the proposed method effectively quantifies the perceptual thresholds and perceptual difference for varied individuals. These findings provide empirical evidence for establishing a more realistic and effective haptic feedback in virtual ophthalmic surgical training.

Virtual surgical assessment evaluates a surgeon’s technical skills by analyzing instrument motion data collected during virtual surgery. Although existing methods classify skills accurately from motion data, they struggle to generalize to unfamiliar surgical patterns because surgeons use varied techniques. To address these limitations, this paper proposes a Geometric Constraints based Bayesian Neural Network (GeomBNN) model for surgical assessment. The geometric constraints enforce orthogonality among nonlinear components of the motion data, enhancing the model’s capacity to capture complex surgical patterns. The Bayesian neural network’s probabilistic framework quantifies predictive uncertainty, enabling adaptation to complex procedures and improving assessment robustness. Experiments on four datasets show that GeomBNN achieves higher classification accuracy than existing methods. What is more, the model exhibits progressive feature attention, dynamically focusing on the most discriminative surgical skill features during training, further validating its effectiveness.

Clinical diagnosis of dermatological diseases often relies on multimodal information, such as dermoscopic images and patient metadata (e.g., gender, medical history). However, most existing interpretability methods primarily focus on a single modality, overlooking the potential interactions between multimodal data, which limits the effective utilization of joint cross-modal features. To address this limitation, we propose a Multimodal Prototype part Network (MPN) that effectively integrates image and metadata, uncovering the underlying relationships between modalities and achieving complementary feature enhancement. Our method captures deep cross-modal interactions through feature fusion and identifies key features via a prototype extraction mechanism, further improving diagnostic interpretability. Additionally, we design a prototype-based loss function to collaboratively optimize the multimodal network. This loss function enhances cross-modal fusion while improving intra-class compactness and inter-class separability, ensuring the accuracy and distinctiveness of the learned prototypes. Extensive experiments on public multimodal dermatological datasets, such as 9, PAD-UFES-20 and ISIC2019, demonstrate that the proposed method outperforms state-of-the-art approaches in terms of accuracy, AUC, and other key metrics. Furthermore, the provided multimodal interpretability effectively supports clinical decision-making, highlighting its potential in advancing dermatological diagnostics.

Accurate and in-time performance monitoring plays a great role in industrial processes. However, since the labeled performances are usually measured by some special devices with a relatively long-time interval, the current deep neural networks for the performance monitoring always treat them only as an output. But actually, in the industrial process like froth flotation, labeled performance at the previous moment could also offer valuable information for performance monitoring at the current moment, especially under unstable working conditions. Therefore, we propose a Siamese time series and difference network (STS-D net), which integrates input features at different time steps and labeled performance at the previous moment effectively. In this article, the proposed STS-D net includes two sub-networks. One is the Siamese time series network, which aims to extract effective and uniform feature representations for the input time series at the current and previous moments; the other is the difference network, which integrates the feature representations of the two input time series with labeled performance at the previous moment to predict the performance at the current moment in an incremental way. Effectiveness of the proposed STS-D net is validated in a real-world froth flotation process.

Mobile Edge Computing (MEC) is a burgeoning paradigm that pushes data and services away from remote clouds to distributed Base Stations (BSs) equipped with MEC servers, which are deployed by Service Providers (SPs) at the edge of cellular networks. Normally, a SP prefers to use its own BSs, instead of those deployed by other SPs, to provide data and storage services. This can not only improve the quality of user experience but also increase its own revenue. In a densely deployed MEC network where a User Equipment (UE) tends to be covered by multiple BSs from varied SPs, how to allocate the resources in the BSs to provide the best service is a challenging problem. In this paper, we propose a novel resource allocation scheme, Decentralized Multi-SP Resource Allocation (DMRA), for densely-deployed MEC networks in order to maximize the total profit of all SPs and provide high-quality services. Our experimental results indicate that the proposed scheme outperforms the existing resource allocation algorithms for MEC.

Online social networks (OSNs) have become effective platforms for influence diffusion. Finding a positive influence dominating set (PIDS) in OSNs can be used to help mitigate social problems such as adolescent drinking and smoking. A set is positive influence dominating if each node in the network is either in the set or has half neighbors in the set. In this article, we propose an efficient greedy algorithm to identify connected PIDS (CPIDS) in large-scale social networks, which utilize two hop coverage information of nodes in the network. Our simulation results show that the proposed approach outperforms existing algorithms in real-world large-scale networks in terms of time cost. Our approach can be potentially used in designing efficient influence diffusion algorithms in OSNs.

The next generation of wireless networks (e.g., Long Term Evolution (LTE)and WiMAX) use multiple techniques to improve channel spectral efficiencies. In this paper we focus on one such technique, namely Coordinated Beam Switching (CBS). With CBS, each sector determines a sequence of beams (possibly with repetitions) over which it continuously cycles. Each sector independently determines its beam pattern with the only constraint being that all sectors use a common cycle period. A beam can be used for the entire frame or one may use different beam patterns for each subband of the frame. Therefore the interference pattern of each subband of a sector repeats with the same period and hence the channel quality can be predicted if we assume slow moving UEs (User Equipment). Such a method may not be suitable for delay sensitive applications but, in the case of VoIP, the periodic arrivals of VoIP packets can be synchronized with the periodic service provided to the UE. Furthermore, VoIP service is limited by UEs at the edge but these are precisely the UEs that benefit from beamforming. In this paper we provide algorithms for accomplishing this synchronization and illustrate the corresponding VoIP capacity gains through simulations.