Impulse attends the 2022 Forum on safe circulation of data elements Online - a special session in the field of government affairs, and helps the construction and innovative development of big data for

7 month 9 Japan , from 《 White paper on the safe circulation of data elements 》 Sponsored by the editorial board 《2022 Forum on safe circulation of data elements — Special session in the field of government affairs 》 Online , Forum around “ government affairs ” Two words , Deeply discuss the cultivation of data factor market and the new situation , How to further ensure the security of government data 、 Promote the innovative development of big data construction of industry government affairs 、 Government data application 、 Government data circulation and other issues , Put forward development suggestions for the construction of digital government .

Zhou yueqian, general manager of impulse online products, attended the forum as a guest , And around “ The cornerstone of high-performance national government data circulation ” Share the theme .

01

Three challenges of data element trading

2020 year 4 month , For the first time, the CPC Central Committee defined data as a new factor of production , Write to 《 Opinions of the CPC Central Committee and the State Council on building a more perfect market-oriented allocation system and mechanism of factors 》 in , It means that the country promotes data nationwide , Especially the open sharing of government data , Encourage data to become an asset , It can be priced at the social level 、 transaction , Put the past 10 The huge data resource value accumulated in the development of the Internet for many years has been fully excavated , Through market-oriented allocation behavior , Let data flow from where it is generated to where it is really needed . Driven by policy , The data element market is booming , But data as a special factor of production , We are facing great challenges in the process of circulation transactions , To be solved urgently .

1、 Data ownership is difficult to define and guarantee
Data is non exclusive and replicable . As the main body of data , Users have the right to control data , But in fact , Data is often not in the hands of users , Whether it's social data , Or consumption data , Are stored in the databases of various service providers , It is regarded as the core asset of Internet companies , Be calculated and analyzed without the user's knowledge , And then make profits .

2、 It is difficult to protect users' data privacy
Even at present, most service providers will make relevant prompts for data flow in the user agreement and obtain user authorization , However, few service providers can adhere to and implement the right to know of users in the whole process . In traditional data circulation , Lack of effectiveness 、 Objective technical means , Users' privacy and trade secrets can be fully protected .

3、 Lack of user incentives
Sustainable development of data market , It is inseparable from the active participation of users . However , At this stage, in most scenarios , The value of users as data subjects has not been reflected , The market also lacks an efficient incentive mechanism to drive users to actively participate in data circulation .

02

“ Blockchain + Privacy computing ” Realize the trusted transaction and circulation of data value

How to solve the three challenges faced by data element trading ？ Zhou yueqian said , Blockchain + Privacy computing can realize the trusted transaction and circulation of data at the value level , Create a good infrastructure for the marketization of data elements .

Privacy computing solves the problem of data privacy information protection , Through hardware 、 Cryptography or distributed means , Protect the privacy of the original data information during the use of data , Implementation data available not visible . Make the data users unable to access the original data , So as to achieve the separation of data use right and ownership .

Blockchain solves the problems of data ownership and user motivation , Through distributed shared database , have “ Don't falsify ”、“ Leave a mark all the way ”、“ It can be traced back to ”、“ Open and transparent ”、“ Collective maintenance ” Other characteristics . Can lay a solid foundation of trust , Create a reliable cooperation mechanism , It has broad application prospects .

Blockchain solves the problems of data ownership and data pricing , By technical means , Clarify the ownership of data assets . The source of data ownership is not clear 、 Data can be copied and tampered with indefinitely 、 The use of data is not unique 、 Data does not have exclusivity and other obstacles to the identification of data ownership . Provide a data asset pricing model for both parties of data transactions . Realize the comprehensive classification and evaluation of raw data , Complex management of evaluation data 、 analysis 、 Storage costs ; Quantify the huge difference in data value generated by different data use methods .

Privacy computing as a new technology , There are still many challenges in the process of commercialization promotion , There are two kinds of ,1） How can privacy computing be widely used in real business 2） How can the security of privacy computing be fully trusted . Zhou yueqian said , Only by solving these two challenges , future , Privacy computing can really assume its due responsibilities in the flow of data elements .

03

High performance 、 Localization

For the two challenges of privacy Computing , Zhou yueqian said , We should build high-performance 、 Localization of the whole stack , Privacy computing technology regulated by national authorities . In the process of achieving this goal , Impulse online has done a lot of work .

The first is technology selection , Impulse online adheres to the basic route of trusted execution environment , The core reason is the trusted execution environment, whether in terms of performance or versatility , Have great advantages . such as , It is more friendly to the knowledge of machine learning , There is no need for large-scale transformation of the original business system , When we move from pilot applications of privacy computing to production level applications , Only the trusted execution environment can support hundreds of millions of levels or even larger data , At the same time, it ensures low performance loss . in addition , Trusted execution environment as a combination of software and hardware , Chip manufacturers can provide trust roots for trusted hardware .

In this trend , Industry focus and standard research learn from the Federation /MPC Transition to a trusted execution environment , meanwhile , In the world , Trusted execution environment TEE As a mainstream solution ,AWS、Google、 Microsoft 、IBM And other giants continue to layout confidential Computing .

With the arrival of the demand for multi-party data exchange 、5G And the development of the Internet of things has brought about a sharp increase in the amount of data , It is bound to consume a lot of computational power , Then the performance of privacy computing will face great challenges , It seems that the multi-path integration of software and hardware based on trusted execution environment is expected to become a way to break the situation . In this context , Impulse online proposed the first hardware in China + Software level optimization ：

1） It is the first in the industry to use GPU Speed up TEE Manufacturers of neural networks to improve training speed ;
2） Impulse data interconnection platform uses Split Learning technology , The model of deep neural network is reasonably split ;
3） The security and privacy capabilities of the trusted execution environment are extended to GPU On the computing chip , Models that support processing more data volumes and data dimensions , It also supports processing machine learning and deep learning models , Greatly improve the calculation efficiency .

04

Self control

Zhou yueqian said , Impulse online is committed to creating a completely autonomous and controllable privacy computing solution combining hardware and software , The scheme is finally presented in the form of an all-in-one machine . A privacy computing platform for data flow built online through impulse , Can schedule different domestic chips , And be able to take advantage of the trust root in domestic chips , And the algorithm library of the state secret algorithm .

up to now , Impulse online has launched, including Zhongke controllable 、 Feiteng 、 Huawei Kunpeng and other software and hardware all-in-one products with different architectures , At different configuration levels 、 Chip level 、 Manufacturers are constantly expanding , Realize the open box availability of privacy computing products for users , For many scenario based algorithms , It can be further optimized at the hardware level , And the whole architecture, from hardware to software to encryption algorithm, can be completely autonomous and controllable .