Introduction to Privacy Computing Concepts and Applications

0, Privacy Computing Background

Policy Background:

In April 2020, the "Opinions of the Central Committee of the Communist Party of China and the State Council on Building a More Perfect Market-Based Allocation System and Mechanism for Factors" was released, using data as a new type of production factor alongside traditional factors such as land, labor, capital, and technology.

In October 2020, the "14th Five-Year Plan for National Economic and Social Development of the People's Republic of China and Outline of Vision 2035" was released, which proposed to accelerate the construction of a digital economy, a digital society, a digital government, a digital China, and a digital economy. New advantages, clarifying the importance of data as a core factor of production.

In the data age, on the one hand, the country needs to build a digital economy and society, and support data open sharing and interconnection; on the other hand, the security issues brought about by data open sharing must also be taken seriously. With the implementation of the "Data Security Law of the People's Republic of China", it not only fills the legal gap in data security, but also greatly promotes the standardization and rapid development of the privacy computing industry.

Market issues:

At present, in the process of data production and processing, data resource aggregation, data circulation and transaction, data model training and deployment in the process of revitalizing the value of data elements , there are still problems such as difficulty in confirming data rights, high input costs, low quality of data sets, and limited data resources . Due to data security issues, data islands exist between data subjects and even within data subjects, and there are many problems in data circulation and data application.

1. The concept of privacy computing

The "Privacy Computing Research Scope and Development Trend" published in 2016 formally proposed the term "privacy computing", and defined privacy computing as: "The computing theory and method for the protection of privacy information throughout the life cycle, is the ownership, management of privacy information. A computable model and axiomatic system of privacy metrics, privacy leakage costs, privacy protection, and privacy analysis complexity when rights and usage rights are separated."

Privacy computing is essentially to solve data service problems such as data circulation and data application on the premise of protecting data privacy.

The philosophy of private computing includes:

"The data is available and invisible, the data does not move the model", "the data is available and invisible, the data is controllable and measurable", "the data is not shared, but the value of the data is shared", etc.

According to the main related technologies of privacy computing technology on the market, it can be divided into three categories: (differential privacy is also included as a data processing method)

Protocol-Based Secure Multi-Party Computation
Federated Learning Based on Modern Cryptography
Hardware-based Trusted Execution Environment

2. Introduction to Privacy Computing Technology

2.1, Federated Learning

Federated learning is a distributed machine learning technology and system, including two or more participants, these participants conduct joint machine learning through a secure algorithm protocol, and can exchange intermediate data without the data of each party being local. form of data, jointly model and provide model inference and prediction services. And the effect of the model obtained in this way is almost the same as that of the traditional central machine learning model.

At present, federated learning technology is relatively mature in traditional machine learning algorithms such as linear regression, decision tree and other models, and the focus of research is on deep learning models.

The use of federated learning technology usually needs to be combined with secure multi-party computing technology, or even blockchain. The development direction of federation technology is to build a unified federation platform to execute data transactions. For a detailed introduction to federated learning, please refer to: Federated Learning Concepts and Applications

2.2, Secure Multiparty Computation

Secure Multi-Party Computation is a technique and system for securely computing agreed functions without the parties sharing their respective data and without a trusted third party. Through secure algorithms and protocols, participants encrypt or convert data in plaintext and then provide it to other parties, and neither party can access the data in plaintext of other parties, thus ensuring the security of data of all parties.

2.3, Trusted Computing

Trusted computing technology based on trusted hardware. Compared with the privacy based on software and protocols, the hardware implementation is more secure and reliable. At present, in China, ants are also doing this.

2.4, Blockchain + Privacy Computing

Blockchain will become an indispensable option in privacy computing products. On the basis of ensuring data credibility, data security, compliance, and reasonable and effective use will be realized. It is mainly reflected in the following three aspects:

Blockchain can guarantee the end-to-end privacy of private computing task data . Through blockchain encryption algorithm technology, users cannot obtain transaction information in the network, and verification nodes can only verify the validity of transactions but cannot obtain specific transaction information, thereby ensuring transaction data privacy, and can be customized according to users, businesses, transaction objects, etc. Data and account privacy protection settings are implemented at different levels to protect the privacy of data to the greatest extent.

Blockchain can guarantee the security of the whole life cycle of data in private computing . The blockchain technology adopts a distributed data storage method. All nodes on the blockchain store a complete piece of data. If any single node wants to modify these data, other nodes can use their own saved backups to falsify them, thus ensuring that Data is not tampered with or deleted indiscriminately. In addition, the asymmetric encryption and hash encryption technology used in the blockchain can effectively ensure data security and prevent leakage.

Blockchain can guarantee the traceability of private computing process . The entire process of data application, authorization, and calculation results is recorded and stored on the chain. The information recorded on the chain can be confirmed by other participants by signing the data to further improve the reliability of the data. At the same time, it can be verified by the hash value. Matching to achieve rapid identification of information tampering. Based on the recording and authentication of data on the chain, smart contracts can be used to associate relevant data on the chain according to unique identifiers to build data traceability. The combination of blockchain and privacy computing enables collaborative computing and data privacy protection between multiple nodes without the need for collection and sharing of original data. At the same time, it can solve the problems of excessive data collection, data privacy protection, and single-point leakage of data storage in the big data mode. Blockchain ensures that the computing process and data are credible, and privacy computing makes data available and invisible. The two combine and complement each other to achieve wider data collaboration.

"Blockchain, because of its technical characteristics such as shared ledgers, smart contracts, and consensus mechanisms, can realize the verification of original data on the chain, and the key data and links of the calculation process. "

3. Privacy computing applications

Government affairs: open sharing of government data, smart city
Finance: credit risk assessment, financial anti-fraud, anti-money laundering, credit investigation, insurance pricing
Medical: joint diagnosis, intelligent consultation, auxiliary medical treatment, pathological analysis
Advertising: Precision Marketing

Government Affairs:

finance:

Medical:

advertise:

【Reference link】

Tencent Privacy Computing White Paper 2021
China Privacy Computing Industry Development Report (2020-2021)

———————————————
Copyright statement: This article is an original article by CSDN blogger "Lin Like", and follows the CC 4.0 BY-SA copyright agreement. Please attach the original source link and this copy for reprinting. statement.
Original link: https://blog.csdn.net/qq_40589204/article/details/116104882