In modern Internet of Things (IoT), visual analysis and predictions are often performed by deep learning models. Salient object detection (SOD) is a fundamental preprocessing for these applications. Executing SOD on the fog devices is a challenging task due to the diversity of data and fog devices. To adopt convolutional neural networks (CNN) on fog-cloud infrastructures for SOD-based applications, we introduce a semisupervised adversarial learning method in this article. The proposed model, named as SaliencyGAN, is empowered by a novel concatenated generative adversarial network (GAN) framework with partially shared parameters. The backbone CNN can be chosen flexibly based on the specific devices and applications. In the meanwhile, our method uses both the labeled and unlabeled data from different problem domains for training. Using multiple popular benchmark datasets, we compared state-of-the-art baseline methods to our SaliencyGAN obtained with 10-100% labeled training data. SaliencyGAN gained performance comparable to the supervised baselines when the percentage of labeled data reached 30%, and outperformed the weakly supervised and unsupervised baselines. Furthermore, our ablation study shows that SaliencyGAN were more robust to the common “mode missing” (or “mode collapse”) issue compared to the selected popular GAN models. The visualized ablation results have proved that SaliencyGAN learned a better estimation of data distributions. To the best of our knowledge, this is the first IoT-oriented semisupervised SOD method.

With the advancement of communication technology and location acquisition technology in the context of modern smart cities, the sharing bike systems offer users the great autonomy and convenience for the last/first-kilometer trip. Meanwhile, we can now able to collect, store, and analyze a large amount of sharing bike data. How to effectively use these massive data to provide better services is an emerging task. However, due to the skewed and imbalanced bike usages for stations located at different places, it is of great significance yet very challenging to predict the potential destinations of each individual trip beforehand so that the service providers can better schedule manual bike re-dispatch in advance. To address this issue, this paper proposes an attention-based deep learning framework for trip destination prediction (AFTER). AFTER first learns the low-dimension representations of users and sharing bike stations via negative sampling strategies. Then, a convolution neural network with an attention mechanism is utilized to predict the future trip destination. Experimental results on a real-world dataset indicate that the proposed framework outperforms several state-of-the-art approaches in terms of precision, recall, and F1.

Helicopters and vehicles are often jointly used to transport key relief supplies and respond to disaster situations when supply nodes are far away from demand nodes or the key roads to affected areas are cut off. Emergency transfer centers (ETCs) are often changed due to secondary disasters and further rescue, so the extant intermodal transportation plan of helicopters and vehicles needs to be adjusted accordingly. Disruption management is used to re-plan emergency intermodal transportation with updated ETCs in this study. The basic idea of disruption management is to minimize the negative impact resulting from unexpected events. To measure the impact of updated ETCs on the extant plan, the authors consider three kinds of rescue participators, that is, supply recipients, rescue drivers, and transport schedulers, whose main concerns are supply arrival time, intermodal routes and transportation capacity, respectively. Based on the measurement, the authors develop a recovery model for minimizing the disturbance caused by the updated ETCs and design an improved genetic algorithm to generate solutions for the recovery model. Numerical experiments verify the effectiveness of this model and algorithm and discern that this disruption management method could produce recovery plans with shorter average waiting times, smaller disturbances for all the supply arrival times, intermodal routes and transportation capacity, and shorter running times. The comparison shows the advantage of this disruption management method over the rescheduling method.

The traditional Internet of Brain Things’ big data information filtering method ignores the extraction of big data features, the filtering effect, and the effect of denoising processing on the simulation results, resulting in low filtering accuracy and poor performance. An effective information filtering and mining algorithm for the Internet of Brain Things based on support vector machine (SVM) is proposed. First, the model construction and feature extraction of the Internet of Brain Things’ big data system are carried out. The correlation feature extraction is performed on the effective information features; the correlation factors of the effective information data are sorted; the main feature quantity of the relevance degree is extracted; and the filter non-association is designed. The information is reasonably filtered; all data are processed, converted to the same interval for processing; data protocol is implemented; and data effective information feature mining is implemented based on the SVM algorithm. The simulation results show that the algorithm is effective for filtering big data and has high precision and superior performance, which shows good application value.