Search

The study of genre preferences of different doctors by leveraging their handled computed tomography (CT) images from multiple hospital databases offers a compelling field of AI-based medical treatment, especially in terms of deciphering and categorizing different diseases and the diseased regions. This research introduces a technique to categorize a large number of medical images into defined groups or “circles” based on common genre preferences, such as “heart diseases” or “osteoarthritis”. Our findings underscore two primary factors: 1) the necessity for an adaptable genre model that can adjust to varied visual characteristics depending on the database, and 2) the inconsistency in the volume of CT images per hospital, with some possessing very few image collection for some particular diseases (for example, a heart hospital has very few joint fracture CT images). To tackle these issues, we propose a regularized latent probabilistic model to depict each doctor’s genre-relevant features as a distribution within a latent manifold space. We subsequently develop a graph that reflects the genre similarities among different doctors. Employing an advanced technique for densely connected graph discovery, we are able to cluster doctors with similar genre preferences into circles. Our experimental outcomes, based on a review of CT image datasets from over 6000 hospitals, validate the effectiveness of our method in precisely distinguishing doctors/users between different genre types. Based on this, various applications can be enhanced.

Image segmentation has demonstrated immense potential in computer vision. In particular, the U-Net architecture, built on fully convolutional networks, is highly suitable for image segmentation tasks. Its encoder-decoder structure effectively captures both local and global features. This approach has achieved remarkable outcomes across various sectors, most notably in medical diagnostics and industrial quality control. However, U-Net, by employing skip connections, fuses different low-level and high-level convolutional features between the encoder and decoder, limiting its ability to effectively integrate useful features and harness contextual information. To address these feature disparities between the encoder and decoder, this paper introduces a novel network structure named Multi-Convolutional Channel Residual Spatial Attention U-Net (MCRSAU-Net). Designed for industrial and medical image segmentation, this model is anchored on the U-Net architecture. It replaces the traditional skip connections with channel attention residual paths featuring multiple convolutions, retaining more low-level features. Moreover, spatial attention module is incorporated in the decoding path to ensure the model concentrates on crucial regions of the input space, enhancing its segmentation capability across varied tasks. The proposed method was subjected to 5-fold cross-validation and testing on three public datasets: Mvtec AD, CHASE DB1, and Kvasir SEG. MCRSAU-Net achieved average Dice coefficients of 0.7755, 0.7651, and 0.8958 for defect segmentation of bottles, woods, and tiles, respectively, with average accuracies reaching 0.9751, 0.9815, and 0.9841. For retinal blood vessel and colon polyp segmentation, it exhibited superior performance, achieving average Dice scores of 0.8540 and 0.7053, and average accuracies of 0.9465 and 0.9195, respectively. These results not only underscore MCRSAU-Net’s strong performance in image segmentation tasks but also demonstrate its significant potential in addressing specific challenges encountered in industrial and medical image segmentation.

In this paper, we propose a permutation index differential chaos shift keying with the noise reduction (NR-PI-DCSK) to alleviate the negative effect of the additive noise and fading interference. Instead of sending a $\beta $ -length chaotic sequence as the reference in the conventional PI-DCSK, the transmitter of NR-PI-DCSK generates $\beta /R$ chaotic samples and repeats $R$ times, which serve as the reference sequence. The permuted copy of the reference is used to carry the data given the predefined permutation index. After the correlation and averaging of these $R$ chaotic segments at the receiver, the variance of noise and fading can be greatly reduced and the performance is improved. Then, we derive the analytical BER expressions for the multipath fading channels, which are verified by the consistent simulations. The results show that the proposed NR-PI-DCSK can achieve up to 3 dB performance gain over the conventional PI-DCSK in different wireless fading channels. In addition, the performance advantage of the proposed NR-PI-DCSK is also demonstrated in a typical UWB channel by simulations. More importantly, the proposed scheme maintains almost the same low transceiver complexity as the conventional ones, since it only requires the correlators. Thus, the NR-PI-DCSK can be seen as a promising candidate for low-cost short-range IoT communications.

To detect large-variance code clones (i.e. clones with many modifications) in large-scale code repositories is difficult because most current tools can only detect almost identical or very similar clones. It has an important impact on downstream software applications such as bug detection, code completion, software analysis, etc. Recently, CCAligner made an attempt to detect the code clones with insertions or deletions in one place, which were called large-gap clones. Our contribution is to develop a novel and effective detection approach of large-variance clones to more general cases for not only the concentrated code modifications but also the scattered code modifications. A detector named LVMapper is proposed, borrowing and changing the approach of sequencing alignment in bioinformatics which can find two similar sequences with more differences. The ability of LVMapper was tested on 8 open source projects datasets, and the results show that LVMapper detected more than 5 times of large-variance clones compared with other state-of-the-art tools including CCAligner. Furthermore, our new tool also presents comparable recall for general Type-1, Type-2 and Type-3 clones with precision of 88.5% on the widely used benchmarking dataset BigCloneBench.

Tailings ponds are a major hazard, and are ranked 18th in the risk assessment of world accident hazards. The saturation line height is one of the most important factors that affects the safety of tailings ponds. Due to the extremely complicated seepage boundary conditions of tailings ponds, a precise calculation method is urgently needed for predicting the saturation lines. Therefore, the dynamic model should be investigated to evaluate the potential for dam breakage. In this paper, based on an analysis of tailings ponds in various regions, we use the long short-term memory (LSTM) algorithm to predict the time-series variation of the saturation line height. To evaluate and validate our model, we compare with traditional models. The results demonstrate that the deep learning method significantly outperforms the traditional methods, provides a new strategy and has significant potential for tailings ponds safety prediction.

Although software defined network (SDN) has been widely recognized for several years, it still has not addressed security issues regarding permission management on SDN controller, one lack of which is the effective permission allocation of application programming interfaces (APIs) for multitenant requirements. Besides, the tendency of integrating SDN and network function virtualization (NFV) introduces a new problem about permission management on both SDN and NFV APIs. This paper presents a fine-grained permission management for secure sharing of APIs in multi-tenant networks. We propose a permission policy language, which provides three-level permission abstraction to define available APIs for multiple tenants to access both OpenFlow switches and network functions. We introduce a permission management framework to effectively enforce permissions and ensure user isolation. A prototype of the proposed framework is implemented on top of the RYU controller. Extensive experiments show that our system is effective for reducing API abuse and only introduces negligible overhead.

Service function chaining (SFC) is able to provide customizable network function services to the traffic flows of different IoT subjects. Nowadays, SFC becomes profound to implement the service requirements of different IoT devices with the flexibility and programmability provided by emerging technologies, software defined network (SDN) and network function virtualization. These techniques play an increasingly important role for service deployment and allow the service requirement for certain IoT device to be specified by different subjects, including SDN applications and network managers. However, independent generation of SFC policies by multiple policy makers over the same device may introduce several problems in the process of deploying SFCs to IoT network. Turning the individual considerations into coherent global SFC policies can be challenging. It requires special process of composition and transition, considering the scenario of combining policies with different concerns specified by different entities who have no insight into the policies of others. In this paper, we propose a composition method to solve the anomalies existing in the process of composing distinct policies in the environment of IoT network with multiple IoT service managers. We design two algorithms for the proposed anomaly-free policy composition method, and implement a prototype. Extensive experiment results show that our proposed method can eliminate the anomalies between policies and only induces trivial overhead in the process of generating data plane rules.