With the development of mobile devices and GPS, plenty of Location-based Services (LBSs) have emerged in these years. LBSs can be applied in a variety of contexts, such as health, entertainment, and personal life. The location based data that contains significant personal information is released for analysing and mining. The privacy information of users can be attacked from the published data. In this paper, we investigate the problem of privacy-preservation of density distribution on mobility data. Different from adding noises into the original data for privacy protection, we devise the Generative Adversarial Networks (GANs) to train the generator and discriminator for generating the privacy-preserved data. We conduct extensive experiments on two real world mobile datasets. It is demonstrated that our method outperforms the differential privacy approach in both data utility and attack error.

The large-scale collection of individuals' mobility data poses serious privacy concerns. Instead of perturbing data by adding noise to the raw location data to preserve privacy of individuals, we propose an approach that achieves privacy-preservation at the statistics level of aggregating mobility datasets with the probabilistic data structure Count-Min Sketch(CMS), which has been widely used to provide efficient statistic functions with a tunable error bound. We use CMS to estimate the population density distributions in the mobility datasets, where the error bound determines utility guarantees. We develop P4Mobi, a novel Probabilistic Privacy-Preserving Publishing framework for Mobility datasets that protects individuals' privacy while complying to a specific utility requirement. We empirically validate the performance of P4Mobi in terms of utility and privacy-preservation by demonstrating its resilience against a recently proposed reconstruction attack model using two real-world datasets. We compare P4Mobi to two state-of-the-art methods and show that with the same level of privacy achieved against our attack model, P4Mobi significantly improves the utility of the published mobility datasets by up to 20%. We also provide a theoretical estimate of the utility achieved by P4Mobi. We found a very consistent match between the estimated and empirical utility of P4Mobi as evaluated on two datasets.

In view of the problem of extrapolating the embedding of a manifold learning from the given data points to incremental data points,the incremental local tangent space alignment algorithm is presented.The method of global coordinates matrix iteration and lower-dimensional coordinates computation is introduced.For incremental learning,the global coordinates of affect points are recomputed,and the updated global coordinates matrix is used to construct global coordinate of the new data points.The lower-dimensional embedding coordinates of the incremental points are formulated by the updated global coordinate matrix and lower-dimensional embedding coordinates of the given data points.The lower-dimensional embedding coordinates of all points are updated by the eigenvalue iteration algorithm.Experiments with the proposed method are carried out to identify the four different mode of the vibration signal of rolling bearing.The experiments result of the incremental points embedding and scatter of the different sorts demonstrates that the incremental local tangent space alignment algorithm can extrapolate the given data points manifold to the incremental learning and the incremental data points can embed effectively to the previously learned manifold.

Diabetic retinopathy (DR), a critical ocular disease that can lead to blindness, demands early and accurate diagnosis to prevent vision loss. Current automated DR diagnosis methods face two core challenges: first, subtle early lesions such as microaneurysms are often missed due to insufficient feature extraction; second, there is a persistent trade-off between model accuracy and efficiency—lightweight architectures often sacrifice precision for real-time performance, while high-accuracy models are computationally expensive and difficult to deploy on resource-constrained edge devices. To address these issues, this study presents a novel deep learning framework integrating depthwise separable convolution and a multi-view attention mechanism (MVAM) for efficient DR diagnosis using retinal images. The framework employs multi-scale feature fusion via parallel 3 × 3 and 5 × 5 convolutions to capture lesions of varying sizes and incorporates Gabor filters to enhance vascular texture and directional lesion modeling, improving sensitivity to early structural abnormalities while reducing computational costs. Experimental results on both the diabetic retinopathy (DR) dataset and ocular disease (OD) dataset demonstrate the superiority of the proposed method: it achieves a high accuracy of 0.9697 on the DR dataset and 0.9669 on the OD dataset, outperforming traditional methods such as CNN_eye, VGG, and UNet by more than 1 percentage point. Moreover, its training time is only half that of U-Net (on DR dataset) and VGG (on OD dataset), highlighting its potential for clinical DR screening.