Privacy-Preserving Federated Deep Learning Diagnostic Method for Multi-Stage Diseases

Diagnosing multi-stage diseases typically requires doctors to consider multiple data sources,including clinical symptoms,physical signs,biochemical test results,imaging findings,pathological examination data,and even genetic data.When applying machine learning modeling to predict and diagnose multi-stage diseases,several challenges need to be addressed.Firstly,the model needs to handle multimodal data,as the data used by doctors for diagnosis includes image data,natural language data,and structured data.Secondly,privacy of patients’data needs to be protected,as these data contain the most sensitive and private information.Lastly,considering the practicality of the model,the computational requirements should not be too high.To address these challenges,this paper proposes a privacy-preserving federated deep learning diagnostic method for multi-stage diseases.This method improves the forward and backward propagation processes of deep neural network modeling algorithms and introduces a homomorphic encryption step to design a federated modeling algorithm without the need for an arbiter.It also utilizes dedicated integrated circuits to implement the hardware Paillier algorithm,providing accelerated support for homomorphic encryption in modeling.Finally,this paper designs and conducts experiments to evaluate the proposed solution.The experimental results show that in privacy-preserving federated deep learning diagnostic modeling,the method in this paper achieves the same modeling performance as ordinary modeling without privacy protection,and has higher modeling speed compared to similar algorithms.

Federation Boosting Tree for Originator Rights Protection

The problem of data island hinders the application of big data in artificial intelligence model training,so researchers propose a federated learning framework.It enables model training without having to centralize all data in a central storage point.In the current horizontal federated learning scheme,each participant gets the final jointly trained model.No solution is proposed for scenarios where participants only provide training data in exchange for benefits,but do not care about the final jointly trained model.Therefore,this paper proposes a newboosted tree algorithm,calledRPBT(the originator Rights Protected federated Boosted Tree algorithm).Compared with the current horizontal federal learning algorithm,each participant will obtain the final jointly trained model.RPBT can guarantee that the local data of the participants will not be leaked,while the final jointly trained model cannot be obtained.It is worth mentioning that,from the perspective of the participants,the scheme uses the batch idea to make the participants participate in the training in random batches.Therefore,this scheme is more suitable for scenarios where a large number of participants are jointly modeling.Furthermore,a small number of participants will not actually participate in the joint training process.Therefore,the proposed scheme is more secure.Theoretical analysis and experimental evaluations show that RPBT is secure,accurate and efficient.

Content Feature Extraction-based Hybrid Recommendation for Mobile Application Services

The number of mobile application services is showing an explosive growth trend,which makes it difficult for users to determine which ones are of interest.Especially,the new mobile application services are emerge continuously,most of them have not be rated when they need to be recommended to users.This is the typical problem of cold start in the field of collaborative filtering recommendation.This problem may makes it difficult for users to locate and acquire the services that they actually want,and the accuracy and novelty of service recommendations are also difficult to satisfy users.To solve this problem,a hybrid recommendation method for mobile application services based on content feature extraction is proposed in this paper.First,the proposed method in this paper extracts service content features through Natural Language Processing technologies such as word segmentation,part-of-speech tagging,and dependency parsing.It improves the accuracy of describing service attributes and the rationality of the method of calculating service similarity.Then,a language representation model called Bidirectional Encoder Representation from Transformers(BERT)is used to vectorize the content feature text,and an improved weighted word mover’s distance algorithm based on Term Frequency-Inverse Document Frequency(TFIDF-WMD)is used to calculate the similarity of mobile application services.Finally,the recommendation process is completed by combining the item-based collaborative filtering recommendation algorithm.The experimental results show that by using the proposed hybrid recommendation method presented in this paper,the cold start problem is alleviated to a certain extent,and the accuracy of the recommendation result has been significantly improved.

Road Distance Computation Using Homomorphic Encryption in Road Networks

Road networks have been used in a wide range of applications to reduces the cost of transportation and improve the quality of related services.The shortest road distance computation has been considered as one of the most fundamental operations of road networks computation.To alleviate privacy concerns about location privacy leaks during road distance computation,it is desirable to have a secure and efficient road distance computation approach.In this paper,we propose two secure road distance computation approaches,which can compute road distance over encrypted data efficiently.An approximate road distance computation approach is designed by using Partially Homomorphic Encryption and road network set embedding.An exact road distance computation is built by using Somewhat Homomorphic Encryption and road network hypercube embedding.We implement our two road distance computation approaches,and evaluate them on the real cityscale road network.Evaluation results show that our approaches are accurate and efficient.