With the support of cloud computing technology, it is easier for financial institutions to obtain more key information about the whole industry chain. However, the massive use of financial data has many potential risks. In order to better cope with this dilemma and better protect the financial privacy of users, we propose a privacy protection model based on cloud computing. The model provides four levels of privacy protection according to the actual needs of users. At the highest level of protection, the server could not access any information about the user and the raw data, nor could it recover the computational characteristics of the data. In addition, due to the universality of the mathematical principle of linear operators, the model could effectively protect and accelerate all models based on linear operations. The final results showed that the method can increase the speed by 10 times, compared with the privacy protection method that only uses local computing power instead of the cloud server. It can also effectively prevent the user’s privacy from being leaked with relatively minimal delay cost, compared with no privacy protection method. Finally, we design a multi-user scheduling model to deploy the model in a real scenario, which could maximise server power and protect user privacy as well.

Achieving high-accuracy chicken face detection is a significant breakthrough for smart poultry agriculture in large-scale farming and precision management. However, the current dataset of chicken faces based on accurate data is scarce, detection models possess low accuracy and slow speed, and the related detection algorithm is ineffective for small object detection. To tackle these problems, an object detection network based on GAN-MAE (generative adversarial network-masked autoencoders) data augmentation is proposed in this paper for detecting chickens of different ages. First, the images were generated using GAN and MAE to augment the dataset. Afterward, CSPDarknet53 was used as the backbone network to enhance the receptive field in the object detection network to detect different sizes of objects in the same image. The 128×128 feature map output was added to three feature map outputs of this paper, thus changing the feature map output of eightfold downsampling to fourfold downsampling, which provided smaller object features for subsequent feature fusion. Secondly, the feature fusion module was improved based on the idea of dense connection. Then the module achieved feature reuse so that the YOLO head classifier could combine features from different levels of feature layers to capture greater classification and detection results. Ultimately, the comparison experiments’ outcomes showed that the mAP (mean average Precision) of the suggested method was up to 0.84, which was 29.2% higher than other networks’, and the detection speed was the same, up to 37 frames per second. Better detection accuracy can be obtained while meeting the actual scenario detection requirements. Additionally, an end-to-end web system was designed to apply the algorithm to practical applications.

Regression analysis is an essential tool for modeling and analyzing data, which can be utilized in various areas for predictive analysis and discovering relationships between variables. However, guidelines such as the model's features, dataset selection, and method settings for using regression models to explore air pollution status are not detailed. This paper applied regression analysis based on air quality data in Beijing from 2017 to 2021, to study the characteristics of regression models, provide research guidance, and update the air pollution research data based on the dataset. This paper drew the latest conclusions: (1) PM2.5 and NO2 are positively correlated on the test set from these five years, yielding a correlation coefficient of 0.7036 by using linear regression. The respective coefficient of determination on small-scale test sets for 2017, 2019, and 2021 is much lower than those derived from a five-year dataset. Single year dataset is not befitting for linear regression analysis. (2) The polynomial regression's coefficient of determination on the training set is higher than that of the linear regression model, which is more proper for regression analysis on a one-year dataset. (3) PM2.5 and NO2 concentrations are strongly positively correlated with whether the air is polluted or not, and the correlation coefficient on the test set from these five years is 0.9697. The accuracy of logistic regression in classifying air pollution status based on these two pollutants' concentrations reaches 0.9430. Besides, this paper proposed some appropriate parameter settings for the logistic regression method provided by Python third-party-library sklearn. Specifically, L2-type regularization is better optimized for the 2017-2021 dataset. L1-type regularization works better when applying a one-year dataset. A boost in the inverse of the regularization strength to 1.8 will optimize the regularization.