Nowadays, more individuals and corporations tend to use machine learning as a service (MLaaS) in cloud computing environment. However, when enjoying the pay-as-you-go mode and flexible capacity of cloud computing, it also increases the risk of privacy leakage for sensitive data. In this paper, we aim to efficiently implement privacy-preserving MLaaS, and focus on k-means clustering over outsourced encrypted cloud databases. Previous works mainly utilize partially homomorphic encryptions, which require a great number of interactive protocols with high computation and communication costs, making them not practical in real-world applications. To better solve this problem, we propose a new secure and efficient outsourced k-means clustering (SEOKC) scheme using fully homomorphic encryption with ciphertext packing technique, which achieves parallel computation without extra cost. The proposed scheme preserves privacy in three aspects: (1) database security, (2) privacy of clustering results and (3) hiding of data access patterns. We provide formal security analysis and evaluate the performance of the proposed scheme through extensive experiments. The experiment results show that our scheme needs much less computation cost (more than three orders of magnitude lower) than the state-of-the-art schemes, and is suitable to be applied on large databases.

The flourishing deep learning on distributed training datasets arouses worry about data privacy. The recent work related to privacy-preserving distributed deep learning is based on the assumption that the server and any learning participant do not collude. Once they collude, the server could decrypt and get data of all learning participants. Moreover, since the private keys of all learning participants are the same, a learning participant must connect to the server via a distinct TLS/SSL secure channel to avoid leaking data to other learning participants. To fix these problems, we propose a privacy-preserving distributed deep learning scheme with the following improvements: (1) no information is leaked to the server even if any learning participant colludes with the server; (2) learning participants do not need different secure channels to communicate with the server; and (3) the deep learning model accuracy is higher. We achieve them by introducing a key transform server and using homomorphic re-encryption in asynchronous stochastic gradient descent applied to deep learning. We show that our scheme adds tolerable communication cost to the deep learning system, but achieves more security properties. The computational cost of learning participants is similar. Overall, our scheme is a more secure and more accurate deep learning scheme for distributed learning participants.

Nowadays, individuals and companies increasingly tend to outsource their databases and further data operations to cloud service provides. However, utilizing the cost-saving advantages of cloud computing brings about the risk of violating database security and user's privacy. In this paper, we focus on the problem of privacy-preserving k-nearest neighbor (kNN) classification, in which a query user (QU) submits an encrypted query point to a cloud server (CS) and asks for the kNN classification labels based on the encrypted cloud database outsourced by a data owner (DO), without disclosing any privacy of DO or QU to CS. Previous secure kNN query schemes either cannot fully achieve required security properties or introduce heavy computation costs, making them not practical in real-world applications. To better solve this problem, we propose a novel efficient privacy-preserving kNN classification protocol over semantically secure hybrid encrypted cloud database using Paillier and ElGamal cryptosystems. The proposed protocol protects both database security and query privacy and also hides data access patterns from CS. We formally analyze the security of our protocol and evaluate the performance through extensive experiments. The experiment results show that the computation cost of our protocol is about two orders of magnitude lower than that of the state-of-the-art protocol while achieving the same security and privacy properties.

The emerging cloud-based IoT paradigm enables IoT devices to directly upload their collected data to the remote cloud and allows data owners to conveniently manage those data through cloud APIs, which has greatly reduced infrastructure investment and data management cost in many IoT applications. Considering that the outsourced data is out of the physical control of data owners and the cloud server cannot always be fully trusted, how to securely delete the unneeded sensitive data stored in cloud to prevent potential data leakage issues is a big challenge. Most of the existing solutions only support coarse-grained deletion and rely on the participation of the cloud server, so their flexibility and practicability are seriously restricted. In this paper, based on an enhanced policy-based puncturable encryption (P-PUN-ENC) primitive, we propose a secure and fine-grained self-controlled outsourced data deletion scheme in cloud-based IoT. The main contribution of our scheme is that it enables data owners to precisely and permanently delete their outsourced IoT-driven data in a policy-based way without relying on the cloud server. To achieve this, we subtly utilize the logical relationship between puncture policy and access policy, and design a policy transform method to convert the puncture process based on puncture policies into update process of access policies. Then, we utilize key delegation technique in attribute-based encryption (ABE) to complete the corresponding key update operations. Additionally, to address the issue of growing key storage and decryption cost in P-PUN-ENC, we propose the outsourced policy-based puncturable encryption (OP-PUN-ENC) primitive by combining the key and decryption outsource technique with P-PUN-ENC. Comprehensive comparisons show that our proposed scheme can better meet the data deletion requirements in cloud-based IoT, and formal security proof and extensive simulation results demonstrate the reliability and efficiency of the proposed scheme.