Recently, smart cities, smart homes, and smart medical systems have challenged the functionality and connectivity of the large-scale Internet of Things (IoT) devices. Thus, with the idea of offloading intensive computing tasks from them to edge nodes (ENs), edge computing emerged to supplement these limited devices. Benefit from this advantage, IoT devices can save more energy and still maintain the quality of the services they should provide. However, computational offload decisions involve federation and complex resource management and should be determined in the real-time face to dynamic workloads and radio environments. Therefore, in this work, we use multiple deep reinforcement learning (DRL) agents deployed on multiple edge nodes to indicate the decisions of the IoT devices. On the other hand, with the aim of making DRL-based decisions feasible and further reducing the transmission costs between the IoT devices and edge nodes, federated learning (FL) is used to train DRL agents in a distributed fashion. The experimental results demonstrate the effectiveness of the decision scheme and federated learning in the dynamic IoT system.INDEX TERMS Federated learning, computation offloading, IoT, edge computing.

a b s t r a c tLaboratory tests were conducted to investigate the effect of wetting-drying (W-D) cycles on the initiation and evolution of cracks in clay layer. Four identical slurry specimens were prepared and subjected to five subsequent W-D cycles. The water evaporation, surface cracks evolution and structure evolution during the W-D cycles were monitored. The effect of W-D cycles on the geometric characteristics of crack patterns was analyzed by image processing. The results show that the desiccation and cracking behaviour was significantly affected by the applied W-D cycles: the measured cracking water content θ c , surface crack ratio R sc and final thickness h f of the specimen increased significantly in the first three W-D cycles and then tended to reach equilibrium; the formed crack patterns after the second W-D cycle were more irregular than that after the first W-D cycle; the increase of surface cracks was accompanied by the decrease of pore volume shrinkage during drying. In addition, it was found that the applied W-D cycles resulted in significant rearrangement of specimen structure: the initially homogeneous and non-aggregated structure was converted to a clear aggregated-structure with obvious inter-aggregate pores after the second W-D cycle; the specimen volume generally increased with increasing cycles due to the aggregation and increased porosity. The image analysis results show that the geometric characteristics of crack pattern were significantly influenced by the W-D cycles, but this influence was reduced after the third cycle. This is consistent with the observations over the experiment, and indicates that the image processing can be used for quantitatively analyzing the W-D cycle dependence of clay desiccation cracking behaviour.

No abstract

When drying a clayey soil, shrinkage and then cracking on soil surface occur due to water loss by evaporation, this phenomenon seems to be temperature dependent. In the present work, experimental tests were conducted on saturated slurry to investigate the desiccation cracking behavior at three temperatures (22, 60 and 105 °C). The initiation and propagation of desiccation cracks during drying was monitored using a digital camera. By applying computer image processing technique, the surface crack ratio (R SC ) which is the ratio of the surface area of cracks to the total surface area of specimen, was defined to quantify crack networks at different water contents. The experimental results show that the initial critical water content (w IC ), which corresponds to the initiation of desiccation crack, increases with temperature rise. After the initiation of a crack, the ratio R SC increases with decreasing water content and then keeps almost constant when the water content becomes lower than the critical water content (w FC ). By comparing the cracking curve with shrinkage curve, it has been found that the cracking curve, to some extent, reflects the shrinkage properties of soil since the w FC is related to the shrinkage limit and slightly influenced by temperature.

Aim: Multiple sclerosis is a severe brain and/or spinal cord disease. It may lead to a wide range of symptoms. Hence, the early diagnosis and treatment is quite important.Method: This study proposed a 14-layer convolutional neural network, combined with three advanced techniques: batch normalization, dropout, and stochastic pooling. The output of the stochastic pooling was obtained via sampling from a multinomial distribution formed from the activations of each pooling region. In addition, we used data augmentation method to enhance the training set. In total 10 runs were implemented with the hold-out randomly set for each run.Results: The results showed that our 14-layer CNN secured a sensitivity of 98.77 ± 0.35%, a specificity of 98.76 ± 0.58%, and an accuracy of 98.77 ± 0.39%.Conclusion: Our results were compared with CNN using maximum pooling and average pooling. The comparison shows stochastic pooling gives better performance than other two pooling methods. Furthermore, we compared our proposed method with six state-of-the-art approaches, including five traditional artificial intelligence methods and one deep learning method. The comparison shows our method is superior to all other six state-of-the-art approaches.

Aim: This paper proposes a novel alcoholism identification approach that can assist radiologists in patient diagnosis. Method: AlexNet was used as the basic transfer learning model. The global learning rate was small, at 10 −4 , and the iteration epoch number was at 10. The learning rate factor of replaced layers was 10 times larger than that of the transferred layers. We tested five different replacement configurations of transfer learning. Results: The experiment shows that the best performance was achieved by replacing the final fully connected layer. Our method yielded a sensitivity of 97.44%± 1.15%, a specificity of 97.41 ± 1.51%, a precision of 97.34 ± 1.49%, an accuracy of 97.42 ± 0.95%, and an F1 score of 97.37 ± 0.97% on the test set. Conclusion: This method can assist radiologists in their routine alcoholism screening of brain magnetic resonance images.

The coronavirus disease 2019 pandemic has posed severe threats to humans and the geoenvironment. The findings of severe acute respiratory syndrome coronavirus 2 (Sars-CoV-2) traces in waste water and the practice of disinfecting outdoor spaces in several cities in the world, which can result into the entry of disinfectants and their by-products into storm drainage systems and their subsequent discharge into rivers and coastal waters, raise the issue of environmental, ecological and public health effects. The aims of the current paper are to investigate the potential of water and waste water to operate as transmission routes for Sars-CoV-2 and the risks of this to public health and the geoenvironment. Additionally, several developing countries are characterised by low water-related disaster resilience and low household water security, with measures for protection of water resources and technologies for clean water and sanitation being substandard or not in place. To mitigate the impact of the pandemic in such cases, practical recommendations are provided herein. The paper calls for the enhancement of research into the migration mechanisms of viruses in various media, as well as in the formation of trihalomethanes and other disinfectant by-products in the geoenvironment, in order to develop robust solutions to combat the effects of the current and future pandemics.

Cerebral micro-bleedings (CMBs) are small chronic brain hemorrhages that have many side effects. For example, CMBs can result in long-term disability, neurologic dysfunction, cognitive impairment and side effects from other medications and treatment. Therefore, it is important and essential to detect CMBs timely and in an early stage for prompt treatment. In this research, because of the limited labeled samples, it is hard to train a classifier to achieve high accuracy. Therefore, we proposed employing Densely connected neural network (DenseNet) as the basic algorithm for transfer learning to detect CMBs. To generate the subsamples for training and test, we used a sliding window to cover the whole original images from left to right and from top to bottom. Based on the central pixel of the subsamples, we could decide the target value. Considering the data imbalance, the cost matrix was also employed. Then, based on the new model, we tested the classification accuracy, and it achieved 97.71%, which provided better performance than the state of art methods.