当前位置：网站首页>2022 love analysis · privacy computing vendor panoramic report | love Analysis Report

2022 love analysis · privacy computing vendor panoramic report | love Analysis Report

2022-06-11 04:04:00 【Love analysis ifenxi】

Report editor

Publicize Love analysis partner & Chief analyst

Hong Yiqun Love analysis Senior analyst

Mengchenjing Love analysis analysts

Catalog

1. Research scope definition

2. Manufacturer Panoramic Map

3. Market definition and manufacturer evaluation

4. List of selected manufacturers

1. Research scope definition

Research scope

Privacy computing , Also known as privacy computing （Privacy-Preserving Computation）, It is based on a set of fusion cryptography 、 Information theory 、 Distributed computing 、 Security hardware 、 Data science and other multidisciplinary technologies , A technical system capable of computing data in an encrypted or non transparent state . Common privacy computing technologies include multi-party secure computing 、 Federal learning 、 Trusted execution environment 、 Homomorphic encryption 、 Differential privacy, etc , By applying privacy computing technology , Enterprise users can provide data privacy protection , Realize data sharing in circulation “ Available not visible ”.

In this report , Love analytics divides the privacy computing market into application layers 、 Platform layer and computing layer . among , The application layer is a pointer to finance 、 government affairs 、 Medical care 、 retail 、 telecom 、 Application solutions including privacy computing products and services provided by business scenarios in various industries such as transportation ; The platform layer refers to the platform products used to support the construction of application solutions , Privacy computing platform ; The computing power layer refers to various computing power solutions for improving the performance of private computing , Including algorithm optimization 、 Hardware acceleration, etc .

Comprehensively consider the concern of the enterprise 、 Industry implementation progress and other factors , Love analysis selected the application layer financial privacy computing solution in this study 、 Privacy computing solutions for government and public services 、 Medical privacy computing solutions , And the platform layer privacy computing platform , common 4 A specific market , Conduct key research .

This report is for the decision-making level of enterprises and government agencies , And big data and artificial intelligence 、 Scientific and technological innovation department 、 Heads of business departments , Through the demand definition of each specific market and the ability to represent the manufacturer , It provides a reference for privacy computing application planning and vendor selection of enterprises and government agencies in various industries .

chart 1： Privacy computing market Panoramic Map

Selection criteria for manufacturers

The manufacturers selected in this report shall meet the following conditions at the same time ：

The manufacturer's products and services meet the manufacturer's capability requirements defined in each market ;
In the past year, the manufacturer has more than a certain number of paying customers （ Refer to the first 3 Each market definition section of this chapter ）;
In the past year, the income of the manufacturer in the specific market has reached the target requirements （ Refer to the first 3 Each market definition section of this chapter ）.

2. Manufacturer Panoramic Map

Love analysis is based on the investigation of Party A's enterprises and typical manufacturers and desktop research , Select selected manufacturers with mature solutions and landing capabilities in the privacy computing market .

3. Market definition and manufacturer evaluation

AI analysis defines the specific market that this privacy computing project focuses on as follows . meanwhile , For some representative manufacturers participating in this report , AI analysis wrote the manufacturer capability evaluation .

3.1 Financial privacy computing solutions

Definition ：

Financial privacy computing solution refers to the solution for banks 、 insurance 、 Data circulation scenarios of financial institutions such as securities , Realize invisible privacy computing products and services for user data , It is mainly used in precision marketing 、 Joint risk control 、 Anti fraud 、 Compliance certification 、 Financial supervision and other scenarios .

End user ：

Bank 、 insurance 、 Big data departments of securities and other financial institutions , Scientific and technological innovation department , Risk control 、 marketing 、 credit 、 Credit Card Center 、 Asset management and other business departments

Core requirements ：

With banks and other financial institutions fully embracing the Internet and digital transformation , Data has become the core element supporting its product service innovation . Although with a broad customer base , Financial institutions have accumulated a large amount of user data , But these data often have the problem of single data dimension . To provide more precision 、 More financial products and services , Financial institutions need to introduce more user behaviors from outside 、 Scene and other data , So as to enrich the data dimension , Extend the application scenario . In the past, it was subject to policy 、 concept 、 Technical and other factors , It is difficult to break through the data security sharing between institutions , Privacy computing technology can protect data privacy , Realize the safe flow of data among institutions 、 Share and apply , Is being widely concerned by financial institutions and began to adopt . The core requirements of financial institutions for privacy computing solutions include ：

Can be in a variety of types , And apply privacy computing technology in highly personalized scenarios . There are many types of data in the financial field , Accordingly , The application scenarios of privacy computing are also very diverse , And every financial institution has certain personalized needs for privacy computing applications , Therefore, financial institutions need to apply a variety of privacy computing technologies , And can integrate different technical solutions in a flexible way . Besides , In some common scenarios , Such as hidden trace inquiry 、 We also need to be able to quickly use standardized solutions in privacy communication .
Strong end-to-end performance in scenarios with high real-time requirements . Some business scenarios of financial institutions , Such as credit approval 、 Transaction monitoring, etc , It is necessary to obtain the calculation results with very low delay , To ensure the quality of customer service , And quickly identify risks 、 Reduce losses . therefore , In such real-time scenarios , Financial institutions need privacy computing solutions with strong end-to-end performance .
By introducing third-party data sources and modeling consulting services of professional institutions , Improve the model effect in a specific scene . The fundamental of financial institutions' application of privacy computing is to improve the business income of products and services , The key to achieving this goal is to build a more effective model in the business scenario , therefore , On the one hand, financial institutions need to introduce appropriate third-party data , Enrich the data dimension of sample data , On the other hand, it is necessary to introduce modeling consulting services of professional institutions , In the data 、 Algorithm selection 、 model training 、 Provide professional guidance on the use of privacy computing tools , So as to improve the effect of the model .
Privacy computing solutions can be rapidly deployed and integrated with the original system at a lower cost . One side , Financial institutions hope that privacy computing applications can be quickly implemented and produce results , Therefore, it is required that the solution can be deployed in a convenient and fast way ; On the other hand , Financial institutions usually have established more complex businesses and IT System , Therefore, the privacy computing solution is required to minimize the transformation of the original system , Integration with the original system .
Meet safety and compliance requirements . The sensitivity of financial data , In addition, the regulatory authorities have multiple requirements for financial data security , Make the financial institutions' privacy computing solutions in data security protection 、 System environment 、 There are high security requirements for the interpretability of the computing process , The supplier's products are required to pass the safety standard evaluation of the authoritative evaluation organization .

Capability requirements of manufacturers ：

Multi party secure computing 、 Federal learning and other privacy computing technology capabilities , And can provide services for users in a more flexible way . One side , Manufacturers need to provide rich operator libraries of encryption algorithms and federated learning algorithm components , Allow users to customize the combination to realize the private computing application for specific application scenarios , Safety in mind 、 performance 、 Different requirements for accuracy . On the other hand , The manufacturer needs to provide a trace query that can be called directly 、 Application solutions such as privacy intersection , Meet the extensive data alignment of financial users in cross agency data collaboration 、ID The need for integration .
Improve end-to-end performance in real-time business scenarios . In real-time computing, network delay is the main factor that will affect the end-to-end performance , Therefore, manufacturers need to focus on optimizing the communication efficiency , For example, by optimizing the process arrangement 、 Task scheduling , Improve the communication efficiency between multiple nodes by improving the parallelism of operators , To improve performance .
It can link rich third-party data resources . Manufacturers need to establish a wide range of data resource ecology , With operators 、 payment 、 Internet 、 The ability to link data resources in government affairs and other fields , Provide more user behaviors for financial institutions 、 Scene and other data . Besides , Manufacturers also need to establish interconnection agreements with other manufacturers , Facilitate financial institutions to call third-party data across platforms .
Provide professional modeling consulting services . Relevant teams of manufacturers need to have rich experience in the financial field , It can provide common algorithms for financial institutions in model construction , And in the data 、 Algorithm selection 、 model training 、 Provide professional advice on the use of privacy computing tools , To achieve better model effect for financial institutions .
Can quickly deploy and integrate privacy computing solutions . In terms of solution deployment , Vendors need to provide agile deployment and delivery methods , For example, the platform adopts cloud native architecture , Supports containerized delivery ; With SDK or API To provide privacy computing power , Support users to quickly build privacy computing applications ; In terms of integration with the original system , Manufacturers need to provide componentized and interfaced services to support financial institutions to transfer and connect data and models between the privacy computing platform and the original system , Reduce the transformation of the original system .
Privacy computing solutions have high security . Manufacturers need to provide perfect data encryption technology 、 Improve the security design of the platform system to improve the security of the solution ; And need to support algorithm flow visualization , And support access to third-party traffic audit tools to verify the data usage to improve the interpretability and reliability of the solution . meanwhile , The manufacturer's products need to obtain the safety standard test of the authoritative evaluation organization .

Included in the standard ：

1. Meet the manufacturer's capability requirements for financial privacy computing solutions ;

2. The number of customers served in this market in the past year 3 Above home ;

3. In the past year, the scale of relevant service revenue in this market is 200 More than ten thousand yuan .

On behalf of the manufacturer assessment ：

（ notes ： The evaluation of the following representative manufacturers is sorted according to the phonetic order of the first word of the manufacturer's abbreviation ）

Insight into technology

About the manufacturer ：

Insight technology is the largest credit industry group in China “ Zhongxin ” incubation 、 National team of Internet information industry “ CETC ” Investment in the leading specialized privacy computing technology service provider , Focus on government affairs 、 Finance 、 Customers in communication and other industries provide privacy computing technology platform construction and Scenario Oriented Data intelligence services . The core members of the company come from zhongchengxin 、 The big Banks 、 And insurance companies , Have rich industry knowledge and service experience .

Product and service introduction ：

Insight technology's core software product insight digital intelligence Federation platform （InsightOne） It is a financial privacy computing platform independently developed by the company , Have scene oriented “MPC+FL” Fusion engine 、 A manageable distributed trust architecture 、 Full computing link privacy security protection 、 Specialized algorithms that go deep into the scene 、 No trusted third party Federation learning 、 Blockchain credit enhancement privacy calculation 、 Multiparty security graph computation and graph Federation learning 、 Cross platform interconnection container and other core technologies . By stealth inquiry 、 Privacy seeking 、 Set operations 、 Construction of functional matrix of joint statistics and joint modeling , Provide credit risk control for users 、 Precision marketing 、 actuarial studies 、 Asset management rating 、 Debt index preparation and other financial scenario application services . Besides , stay InsightOne Based on software services , Insight also developed fusion computing 、 The Internet 、 Privacy computing high-performance all-in-one information and innovation machine products for storing hardware resources InsightStation, It can satisfy the independent control of financial and government enterprise customers 、 Out of the box requirements .

Vendor evaluation ：

Insight into the commonality and security of technology's privacy computing products and services on the platform 、 Scenario service capability for users in the financial industry 、 To the extent that the business effect of financial institutions is improved , And cross platform collaboration .

Insight into technology's privacy computing platform InsightOne It has high versatility and flexibility .InsightOne The platform adopts the fusion engine architecture for computing scenarios , Multi party secure computing 、 Algorithms such as federated learning are split into refined operators , Combined with differential privacy 、 Homomorphic encryption 、 Zero knowledge proof and other technologies , Users can flexibly combine the underlying operators according to their needs , Integrate multiple technologies and learn from each other , Form a calculation process for specific needs , So as to meet customers' requirements for functions in different computing scenarios 、 performance 、 Security 、 Different requirements for calculation accuracy .

For the needs of users in the financial industry , Insight technology has scene service capabilities including data link and business modeling . For users who need multi-party data fusion applications , Insight into technology in government 、 Operator, 、 Electric power 、 Internet 、 A large number of compliance data owners such as credit bureau have deployed privacy computing nodes , Have the ability to link rich data resources , Combined with privacy computing technology , It can continuously empower downstream users in the financial industry . meanwhile , Insight technology is based on its long-term accumulation of technology and services in the financial field , It also provides business modeling and consulting services for users , In its InsightOne Dozens of banks are built into the platform 、 insurance 、 negotiable securities 、 Algorithms commonly used in industries such as asset management , It can be better used for scenario application modeling . for example , Insight technology builds a privacy computing platform for Bohai bank , Based on the platform capability, the performance data of credit card users in the bank can be more securely shared with external operators 、 E-commerce and other multi-party data association , Build the installment marketing model of credit card bills , Serve the development of variance dissimilation marketing strategy , Enhance user experience .

InsightOne The platform can be well compatible with users' original technology stack under the condition of small transformation of the original system of financial institutions . One side , The privacy computing algorithm model is integrated with the original risk control of financial institutions through a distributed engine 、 Marketing and other business scenario models have been deeply optimized and integrated ; On the other hand , Through componentized and interfaced Services , Transfer and connect data and models between the privacy computing platform and users' original systems . therefore ,InsightOne The platform can be used without changing the user's original system habits , Maximize compatibility with different technology stacks .

Insight technology also actively explores cross platform connectivity . In terms of technological innovation , Insight technology has been developed “ Resource containers + Algorithm container + Primitive container ” Three layer container technology , That is to provide customers with an example of “ patch board ” The same privacy computing base , Realize the compatibility and interoperability of algorithm plug-ins from different manufacturers , It breaks the rules of different privacy computing vendors “ Calculate the island ”; In terms of standard setting , Insight technology has actively led and participated in IEEE、CCSA、TC601、TC260 A series of technical standards for cross platform interconnection of privacy computing formulated by institutions such as ; In application practice , Insight into technology and ant group 、 Weiwei technology has realized the first fully peer-to-peer algorithm protocol interworking between multi-party heterogeneous privacy computing platforms in the industry , Under the leadership of the head office of China Merchants Bank , The first large-scale joint-stock commercial bank interconnection platform in China has been built .

In terms of product safety ,InsightOne The platform has passed the multi-party security computing and federal learning functions of the Chinese Academy of communications 、 performance 、 Security 、 Evaluation of a full range of privacy computing products such as blockchain assistance , As well as the national financial technology evaluation center multi-party security computing and federal learning financial application dual evaluation . Its independent and self-developed Federation computing framework without trusted third party , Through interactive learning and distributed learning between peer-to-peer networks , It solves the risk problems of relying on third-party parameter operators and key distribution in the open source federated computing framework .

Typical customers ：

China Merchants Bank 、 Bank of Beijing 、 Huaxia bank 、 Bohai bank 、 China life,

Rich digital technology

About the manufacturer ：

Fosun technology was founded in 2016 year , It is one of the leading privacy and security computing technology service providers in China , Focus on federal learning 、 secure multi-party computation 、 In the field of encrypted computing, such as stealth query , The business scenario is based on finance 、 Operator, 、 Give priority to government affairs , And expand to medical 、 Judicial supervision 、 Industrial interconnection and other fields . Fushu technology is the leading unit of the first national standard for privacy computing interconnection protocol , Deeply participate in the information security and Standardization Committee 、 Gold standard committee 、 Formulation of standards such as the Ministry of industry and information technology .

Product and service introduction ：

The core product of Fusu technology is Avatar Secure computing platform , The platform includes federal learning 、 secure multi-party computation 、 Trace inquiry 、 Open platform and other modules , It can provide joint statistics for customers in finance and other industries 、 Joint modeling 、 Functions such as hidden trace query . among ,FMPC The open platform brings the privacy computing power of rich digital technology to SDK and API Provide to users in the form of , It is convenient for users to quickly build privacy computing solutions , And adapt and interconnect with the platforms of other users .

stay Avatar Based on the secure computing platform , Fusu technology also provides users with modeling and consulting services for business scenarios in financial and other industries .

Vendor evaluation ：

Rich digital privacy computing products and services , In terms of the integrity of platform functions 、 Flexibility in how you use it 、 Platform openness and security , And modeling and consulting services for users in the financial industry .

Rich digital technology Avatar The secure computing platform has perfect functions , And the use is flexible .Avatar The secure computing platform incorporates federated learning 、 secure multi-party computation 、 Zero knowledge proof and other privacy computing technologies , Users can call the rich operators and functional modules in the platform in the visual operation interface for combination , Form the required application solutions . meanwhile , The platform also provides operation management components 、 System security components 、 Compliance audit component 、 Infrastructure layer 、 Service layer and other functional components , Let users realize out of the box 、 Rapid deployment .

Avatar The security computing platform has high openness . One side ,FMPC The open platform combines the multi-party secure computing and federated learning capabilities of rich number technology SDK and API Provide to users in the form of , Users can quickly build privacy computing application solutions . On the other hand ,FMPC The open platform supports users to access data through standard protocol interfaces 、 Algorithm 、 Models and other resources are interconnected , Help users realize multi-party data fusion and federated learning modeling . meanwhile , Fusu technology actively promotes the interconnection work in the industry , And take the lead in formulating the first national standard for privacy computing interconnection .

Fusu technology focuses on providing professional modeling and consulting services for users in the financial industry , To achieve better model effect . In terms of data docking , Rich digital technology has accumulated a wealth of operators 、 payment 、 Internet devices and other data source resources , meanwhile , Based on a deep understanding of the data , Rich data technology can provide effective data sources for financial users in modeling 、 And suggestions on data field selection ; In terms of model building , The Fusu technology team has large financial institutions 、 Rich experience in fintech companies , Business scenarios in the financial field 、 Professional knowledge and deep understanding , Combined with rich experience in algorithms and knowledge of privacy computing tools , It can help users in the financial industry pull new customers 、 High net worth customer identification 、 Risk control anti fraud 、 Provide professional consulting services in the construction of various models such as credit enhancement evaluation , Improve model effect .

In terms of platform security ,Avatar Secure computing platforms focus more on security , While giving consideration to performance , Ensure data encryption range and encryption strength , To improve platform security . meanwhile , Rich digital technology has a set of systematic methodology and tools , From system security 、 Algorithm security 、 To achieve security and other aspects to provide security guarantee for the platform , Through the visualization of privacy computing process, users can understand the logic and data flow of each link , Avoid black box operation , Increase user security . Besides , Avatar The security computing platform has passed the special security evaluation of the multi-party security computing products of the ICT Institute .

Typical customers ：

China UnionPay 、 The bank of China, 、 Bank of Communications 、 China Merchants Bank 、 Guangdong Rural Credit Union

Weiwei Technology

About the manufacturer ：

Weiwei technology was founded in 2019 year , Is a service provider focusing on privacy protection computing technology , The founding team is from the University of California, San Diego UCSD Etc , With deep privacy Computing 、 Academic and practical experience in the field of biomedical information , Most of the team members come from IBM、Google、Thermo And so on . Weiwei technology has developed a complete set of independent 、 Security 、 Controllable privacy protection computing platform products , Business scenarios cover medical 、 Finance 、 insurance 、 government affairs 、 Security, etc .

Product and service introduction ：

Weixin privacy protection computing platform of Weiwei technology has independent intellectual property rights , Has passed many authoritative evaluations , Security 、 Controllable privacy computing infrastructure platform . The platform includes a secure federated learning system 、 Multiparty secure computing MPC System 、 Super fusion sandbox （TEE） Three core modules , It can provide users with privacy query 、 Privacy modeling / analysis 、 Privacy reasoning and other privacy computing functions , And performance 、 Security 、 Accuracy and other requirements .

For users in the financial industry , Weiwei technology provides a computing platform for privacy protection of financial services (NovaFintech). As a sub platform of Weixin privacy protection computing platform , It can meet the needs of users in the financial industry 、 Analysis function 、 Use the differentiated requirements of interaction . Provides data across multiple sources （ Far greater than 3 Fang ） Joint privacy intersection and privacy modeling capabilities , It can also support malicious models 、 Prevent side channel attacks and other security functions . At the same time, it provides differentiated data source link resources , Meet the needs of specific financial scenarios .

meanwhile , Kuiwei technology also provides products based on localization CPU、 Accelerator card 、 Wei Xin all-in-one machine product of trusted execution environment , Provide users with privacy protection computing solutions combining software and hardware .

Vendor evaluation ：

Weiwei technology's privacy computing products and services are facing users in the financial industry , Support rich application scenarios 、 High performance and high precision for complex application scenarios 、 Multiple data source links 、 And platform security .

The privacy protection computing platform of Weiwei technology can support the rich application scenarios of users in the financial industry . Weixin financial services privacy protection computing platform (NovaFintech) It is a privacy computing sub platform for financial scenarios . The platform is based on the common privacy protection computing platform base of Weiwei technology , Through the flexible deployment of upper layer applications , So as to solve the privacy computing needs of financial customers in different scenarios .NovaFintech Help bancassurance institutions conduct joint analysis and modeling of security with the help of industry wide data , Has expanded its business in the financial industry 、 Digital marketing 、 Accurate customer acquisition 、 Smart risk control 、 Intelligent anti fraud and other specific scenarios continue to land .

The privacy protection computing platform of Weiwei technology has the advantages of performance and accuracy in processing complex financial business scenarios . The privacy protection computing technology of Weiwei technology was first based on the continuous iterative optimization of unstructured data in complex scenes in the medical field , Therefore, the platform has obvious performance and accuracy advantages in complex financial scenarios . for example , In the process of cooperation with a certain financial institution , It is necessary to calculate and process the privacy protection of the face recognition system , Because the complexity of face image data is much higher than that of structured data , The traditional multi-party security learning technology can not meet the requirements of accuracy and performance . Weiwei technology combines the machine learning technology model with the self-developed privacy protection computing algorithm , While ensuring the high accuracy of the face recognition system , The privacy computing performance is guaranteed to the greatest extent .

Weiwei technology provides financial customers with data connectivity based on privacy computing by connecting rich data sources . Weiwei technology in communication operators 、 Bank 、 UnionPay 、 Public security departments and other institutions , Multiple private computing nodes are deployed , Strengthen the construction of privacy computing ecological network , Continue to empower downstream users in the financial industry . For example, during the construction of the anti gambling, anti fraud and anti money laundering platform , Weiwei technology extracts features from multi-source data , Model design and strategy matching , For transaction process data 、IP data 、 Comprehensive analysis of card data , Qualitative analysis of funds and accounts , To gamble 、 Fraud involved 、 Virtual currency 、 The money laundering account shall be fully supervised , And provide early warning services to relevant institutions .

In terms of product safety , Weiwei privacy protection computing platform has passed the special evaluation of the basic ability of the federal learning and trusted execution environment of the Academy of information and communications , As well as the product evaluation of the privacy protection computing platform of the Ministry of public security and the three-level filing and evaluation of the security classification protection of the national information system .

Tongdun technology

About the manufacturer ：

Tongdun technology is a leading AI technology enterprise in China , Focus on decision intelligence , Through the decision intelligence platform based on artificial intelligence and the shared intelligence platform based on privacy Computing , Focus on financial risks 、 Security and digitalization of government and enterprises , Using the company's algorithms 、 Tools and data ecology , Help customers guard against fraud and security risks , Promote the intelligent decision-making process , Improve the flexibility of business decisions 、 Agility and accuracy .

Product and service introduction ：

Zhibang platform is the privacy computing platform of tongdun technology , The platform is based on Federated learning 、 Multiparty secure computing 、 Privacy security intersection and other privacy computing technologies , Combined with the industrial operator Library Caffeine、 Federal data security exchange protocol FLEX、 Computing and communication engine Ionic And other functional components , For customers in precision marketing 、 Risk control joint modeling 、 Anti electronic fraud 、 Anti money laundering and other scenarios to achieve privacy computing applications .

meanwhile , Zhibang platform has built a knowledge Federation ecosystem from data to knowledge transformation , It can realize knowledge sharing , And use the data of all participants for joint calculation 、 Joint modeling 、 Federal forecast , And the use of knowledge networks for knowledge reasoning 、 Knowledge deduction , So that knowledge can flow freely between different knowledge sources , Support intelligent decision-making .

Vendor evaluation ：

The privacy computing products and services of tongdun technology can be rapidly deployed and achieve business results , Support intelligent decision-making based on knowledge network , It can provide rich data sources 、 Application scenarios and modeling consulting , And higher performance and security .

The Zhibang platform provides the whole process of privacy computing solutions , Be able to quickly deploy and achieve business results . In terms of platform architecture , The platform adopts k8s Cloud native layered architecture , It can seamlessly connect users' private cloud 、 Public cloud 、 Mixed cloud and other environments , And support container based agile delivery and rapid release ; In terms of how privacy computing is used , The industrial verified operator library developed by tongdun technology includes typical operators and protocols for privacy computing , Users can flexibly and autonomously build privacy computing workflows , You can also directly call the encapsulated operator combinations in the platform , Lower the use threshold ; In terms of application services , The platform connects with tongdun technology or third-party marketing at the upper layer 、 Risk control, etc SaaS application , It can speed up the application landing ; Besides , The platform also provides data exchange sandbox function , It is difficult to access multi-source heterogeneous data , And inconsistent data standards .

Zhibang platform can further transform data into knowledge based on the knowledge Federation framework system of tongdun technology , To realize intelligent decision-making in various business scenarios . The framework of knowledge Federation starts from the information layer through a series of tool components of the platform 、 The model layer 、 The cognitive layer 、 The four levels of knowledge layer realize the transformation from data to knowledge . At the information layer , The platform cleans the data of all participants 、 Conversion and encryption ; At the model level , The platform aggregates and updates the model parameters of all parties involved in training through federated learning ; At the cognitive level , The platform passes through the full connection layer in the neural network 、 After feature extraction, nested features such as high-level semantic features further update the model , Improve its accuracy ; At the knowledge level , The platform builds knowledge network , And knowledge fusion 、 Knowledge reasoning , Make intelligent decisions .

The huge customer ecosystem accumulated by tongdun technology can provide rich application scenarios and professional modeling consulting services for customers in the financial and other industries . Tongdun technology has served more than 10000 customers , Customer types cover 22 Industries ,118 Segments , One side , It can provide customers with a wide range of data sources , Support precision marketing 、 Intelligent risk control 、 Anti Telecom fraud 、 Anti money laundering and other financial business scenarios ; On the other hand , It can provide a wide range of application scenarios for the implementation of privacy computing technology , And provide professional modeling consulting services . Besides , Tongdun technology is interconnected with China Merchants Bank Huidian platform and systems among friends through Zhibang products , And open its self-developed data exchange protocol FLEX Or compatible with other interconnection protocols , Build a knowledge sharing ecosystem for customers .

Zhibang platform has high computing performance , It can meet the real-time requirements of users in finance and other industries in specific scenarios . The platform makes in-depth customization and Optimization for memory data fragmentation under massive data , It can complete the secure alignment of billion level data in hours ; meanwhile , Communication framework developed by tongdun technology Ionic By optimizing the process layout 、 Task scheduling , Improving the parallelism of operators improves the communication efficiency of Federated algorithms , Therefore, the modeling and calculation efficiency of the platform is improved 9 times 、 Resource consumption is reduced 70%-80%、 The model accuracy and local error are within 0.1%; Besides , The platform also supports distributed online prediction , Single machine performance can reach above thousand qps. Based on these performance improvements , The platform can be used for credit approval 、 In the scenarios with high real-time requirements such as anti fraud, it meets the following requirements: 200 Performance requirements for millisecond response .

In terms of safety and compliance , Zhibang platform adopts the highest security level of inadvertent transmission 、 Homomorphic encryption algorithm to build privacy computing applications . meanwhile , The platform can be compatible with reliable third-party traffic audit tools , Such as Tcpdump, Verify the purpose of each piece of data , Timely discover the risks such as disclosure of plaintext , Ensure that the calculation results meet the privacy calculation requirements . Besides , Zhibang platform has also been based on multi-party security computing through the ICT Institute , And the special evaluation of basic ability of data circulation products based on federal learning .

Typical customers ：

Industrial and Commercial Bank of China , China Merchants Bank , Orange family

3.2 Privacy computing solutions for government and public services

Definition ：

Privacy computing solutions for government and public services refer to data flow scenarios for government and public services , Realize invisible privacy computing products and services for user data . It is mainly used in various data sharing and data opening scenarios of government departments .

End user ：

Local big data bureaus 、 Data exchange 、 The information departments of the government and public service institutions such as the Commission, office and bureau 、 Big data Department 、 Business handling department, etc

Core requirements ：

With the marketization of data elements 、 The government enterprise data fusion and other policies have been issued one after another , In recent years, government agencies such as government and public services are actively promoting the internal sharing of government data among different departments , So as to improve the level of government governance and public service efficiency , meanwhile , The further opening of government data to the outside world can also empower many industries and enterprises , Improve social and economic benefits . Because the government data includes social security 、 Accumulation fund 、 Taxation 、 traffic 、 Hydropower and other data , Scattered in different departments , In addition, government data involve a large number of citizens' privacy , Strict control and other factors , In the past, the government data circulation and sharing approval procedures were cumbersome , Coordination is very difficult . Privacy computing technology can protect data privacy , Realize the safe flow of data among institutions 、 Share and apply , It is being tried by government agencies . The core requirements of government departments for privacy computing solutions include ：

Can use privacy computing technology to get through multi-party data , And the threshold of privacy computing is low . At present, the main demand of government departments for privacy computing applications is to get through multi-party data , Meet the needs of common business scenarios , Such as statistical analysis 、 Joint modeling 、 Data query, etc , Therefore, government departments need to use multi-party security computing 、 Technologies such as federated learning enable these applications . Besides , The government privacy computing solution has a large number of users in business departments , Therefore, it is necessary to call encapsulated applications directly in common scenarios .
Ensure good computing performance in multi-party application scenarios . Government privacy computing applications usually involve data sources of multiple government departments , And with the further opening of government data , There may also be dozens or even hundreds of data applications . therefore , Government departments need to ensure high computing performance in such scenarios .
It can make data elements fully circulate and open to the outside world in a safe and reliable way . The premise for government data to exert collaborative value through privacy computing is that the data can cross departments 、 Across the region 、 Full circulation and openness across industries , therefore , Government departments need a perfect data operation system to support the circulation and opening of data , And confirm the right of data through blockchain technology , Realize data capitalization .
Meet safety and compliance requirements . Government data involves a large number of citizens' privacy , Government departments strictly control them , Therefore, government departments need privacy computing solutions in data security protection 、 The system environment has high security . meanwhile , To ensure the autonomy and controllability of the core technology of privacy computing , Government departments require that the core models of software and hardware related to privacy computing solutions be fully localized . Besides , The government affairs department also requires the supplier's products to pass the safety standard evaluation of the authoritative evaluation organization .

Capability requirements of manufacturers ：

Multi party secure computing 、 Federal learning 、 Homomorphic encryption and other privacy computing technologies , And lower threshold for business users . One side , Manufacturers need to provide multi-party secure computing 、 Statistical analysis of government departments supported by technologies such as federal learning 、 Joint modeling 、 Data query and other application requirements , And manufacturers especially need high-performance homomorphic encryption technology , To improve computing performance in multi-party computing scenarios , And reduce the transformation of the business system of the government departments . On the other hand , Manufacturers need to customize privacy computing applications for some scenarios commonly used by government departments , And it can be called by business users in a visual way , Or in the form of an all-in-one software and hardware machine .
Improve computing performance in multi-party application scenarios . One side , Vendors need to target joint Computing 、 Scenarios such as stealth query improve computing performance , Including algorithm optimization 、 Communication optimization 、 Hardware acceleration and other performance optimization ; On the other hand , Manufacturers need to further provide multi tenant management capabilities , Cut computing resources , Improve the computing performance under high concurrency and multi tenancy models .
Provide data operation services based on blockchain technology . The manufacturer shall provide data circulation / Sharing platform , And build a data operation service system centered on the platform , Connect multi-party data source agencies and data application agencies , To support all parties to complete data circulation and sharing on the platform . meanwhile , Manufacturers need to apply blockchain technology in the platform , By looking at the data 、 Data processing paths and rules 、 Party identity 、 Distribution mechanism, etc. for chain deposit , Ensure the safety and credibility of data and participants in all links of data circulation .
Privacy computing solutions have high security . Manufacturers need to provide perfect data encryption technology 、 Improve the security design of the platform system to improve the security of the solution ; meanwhile , Manufacturers need to realize the localization of all core software and hardware models related to privacy computing ; Besides , The manufacturer's products need to obtain the safety standard test of the authoritative evaluation organization .

Included in the standard ：

1. Meet the capability requirements of manufacturers of government and public service privacy computing solutions ;

2. The number of customers served in this market in the past year 3 Above home ;

3． In the past year, the scale of relevant service revenue in this market is 200 More than ten thousand yuan .

On behalf of the manufacturer assessment ：

（ notes ： The evaluation of the following representative manufacturers is sorted according to the phonetic order of the first word of the manufacturer's abbreviation ）

Zero Technology

About the manufacturer ：

Zero technology was founded in 2016 year 8 month , It is a national high-tech enterprise with leading blockchain underlying technology and in-depth application scenarios . The company aims to use blockchain and privacy computing technology , Create data value circulation infrastructure , To ensure the reliable and orderly flow and safe and efficient application of data among multiple agents , Serve the car 、 Finance 、 government affairs 、 Double carbon 、 Culture and other fields are deeply digitized .

Product and service introduction ：

The zero privacy computing service platform is based on multi-party secure computing 、 Federal learning 、 Homomorphic encryption and other technologies , Provide users with multi-party data fusion 、 model training 、 The whole process from model evaluation to application deployment . meanwhile , The platform provides joint modeling 、 Hidden query 、 Security matching 、 Independent functions or service components such as security statistics , Government data can serve finance safely and efficiently 、 Double carbon and other scenarios , And implement privacy computing applications .

Help users in government affairs and other industries to realize the privacy and security management of the whole life cycle of data elements , Zero technology combines privacy computing and blockchain technology , Launched the zero data circulation service platform , On the basis of providing complete privacy computing platform functions , By applying blockchain Technology , Keep the operation and processing records of the whole process of data circulation in the chain , And use smart contracts on the chain to manage the computing process , Ensure the safety and compliance of data use .

Vendor evaluation ：

The privacy computing products and services of zero technology are flexible 、 It can combine blockchain technology to ensure the credibility of data and data sharing participants 、 Higher performance and security .

The privacy computing function of the zero privacy computing service platform has high flexibility , It can realize a variety of privacy computing applications , And take into account the different preferences of users for performance and security in different scenarios . Zero technology decouples the federated learning and multi-party secure computing engines in the platform privacy computing engine system into operator libraries , among , The federated modeling component library contains data fusion 、 Data preprocessing 、 Feature Engineering 、 Linear regression 、 Logical regression 、XGBoost、K-Means、 Algorithm components such as model reports ; The joint computation operator library contains four operations 、 Logical operations 、 Operators with different computing power, such as statistical computation , Users can combine different operators to build the required privacy computing model through the visual interface . meanwhile , The privacy computing engine system of the platform also includes hidden queries for specific application scenarios 、 Security matching engine , Users can call their functional models directly , Implement related applications .

The zero data circulation service platform is a product combining privacy computing and blockchain , It can guarantee the credibility of data and the identity of data sharing participants , Improve the efficiency of data sharing and circulation . One side , With blockchain Technology , The platform can collect data 、 The path and rules of data processing are linked and stored , And open it to the third party through the data sharing directory of the platform , So as to ensure the compliance of data in the whole life cycle of circulation 、 trusted , And support the post audit traceability ; On the other hand , The platform can realize decentralized private computing scheduling with the help of blockchain technology , When initiating a privacy computing task , The platform can generate smart contracts on the chain as scheduling instructions , The identity of the parties is stipulated by the contract , And the data or calculation rules that each participant needs to provide , Thus, the scheduling instruction is prevented from being tampered with . Besides , The platform can further set the allocation mechanism recognized by all participants through smart contracts , Evaluate and identify the contribution of each participant , And distribute the income based on the contribution . For example, in “ Blockchain + Park energy double carbon ” Application scenarios , The evaluation of the green degree of an enterprise needs multi-dimensional data , Including water for enterprises 、 electric 、 gas 、 Thermal energy , It also includes the industry of the enterprise 、 scale 、 Operating income 、 Tax payment and other indicators , It also includes the enterprise's green investment in photovoltaic, etc . The traditional Internet filling mode is inefficient in collecting data , And it is easy to disclose enterprise privacy , Factors such as adverse impact on the enterprise , Therefore, enterprises are often reluctant to fill in . Through the zero data circulation service platform , It can realize the privacy intersection and security fusion of multi-party data , And the relevant characteristics of Federated learning and model algorithm are used , Train and form an enterprise green credit rating model . Through the built-in comprehensive empirical logic model of enterprise green score and financial score , That is, according to the weight setting of each index of the model , It can effectively evaluate the green credit status of enterprises , Finally, the green credit rating results of enterprises in the park will be output . In the whole process, the data provider will not disclose data by using privacy computing , The platform does not disclose the calculation model , Blockchain solves the trust problem , Ensure that critical data is not tampered with 、 The calculation process is transparent and traceable .

The zero privacy computing service platform has high computing performance . Zero technology has greatly optimized the algorithms built in the platform , By reducing the number of Communications , Through the use of more low-level development language to implement the basic algorithm operators to improve the operation efficiency , Optimize computing logic and improve computing performance . meanwhile , Combined with hardware accelerator card , And the optimization of calculation and scheduling process , The platform can be used in some privacy computing scenarios , For example, in the hidden query of 100 million level data , More efficient than similar solutions 25%.

On the security of privacy Computing , Zero technology from the system 、 Algorithmic protocol 、 data 、 application 、 signal communication 、 Password and other levels have done a lot of permission control 、 Encryption processing, etc , It can guarantee the security requirements of users . meanwhile , The zero federal learning platform has also passed the special evaluation of the federal learning basic ability of the Academy of information technology , In terms of dispatching management capability 、 Data processing power 、 Algorithm implementation 、 Effect and performance 、 Security and so on , It has been recognized by authorities .

Typical customers ：

A district big data Bureau in Qingdao 、 Big data Bureau of a city in Jiangsu 、 A big data center in Shanghai

Homomorphic technology

About the manufacturer ：

Homomorphic technology is a privacy computing data protection service provider , Homomorphic encryption technology based on self-control , In government affairs 、 Finance 、 Military civilian integration and other fields provide users with data exchange, sharing and privacy computing data protection services , Implementation data available not visible 、 Standardization of compliance data 、 The whole process of data application is controllable .

Product and service introduction ：

“ Privacy computing application integration platform ” It is the main privacy computing platform product of homomorphic technology , Based on self-developed super high speed homomorphic encryption algorithm , The privacy computing application integration platform provides users with data privacy protection and data sharing application capabilities , Can be used for machine learning 、 Data statistics 、 Privacy query and other scenarios . meanwhile , Users can directly use the preset privacy computing application in the operating system , Or customize the upper layer privacy computing application based on the development interface , Enabling federated learning and multi-party secure computing as the underlying infrastructure .

The privacy computing all-in-one machine of homomorphic technology is the world's first ultra-high speed homomorphic encryption machine , Integrate SM2、SM3、SM4 And high-speed homomorphic encryption algorithm , Provide standardized data output capability for data sharing and privacy computing , Implement intrusion free privacy computing solutions . High speed chip with hardware algorithm , Provide privacy computing hardware capabilities , Rely on the standard commercial secret hardware platform , Provide ultra-high speed homomorphic encryption capability and a variety of development interfaces , Enable data privacy protection . meanwhile , Provide security authorization management services , Provide independent device keys for each client user , Ensure the safety authorization and use of the upper layer and equipment .

Vendor evaluation ：

The application of homomorphic technology privacy computing products and services to government industry users , Have the ability to meet the needs of users with different technical levels 、 Small business system transformation 、 It can guarantee performance in a multi-party privacy computing scenario 、 And higher security .

Homomorphic privacy computing application integration platform can meet the needs of users with different technical levels , And ensure a good interactive experience . Business users can directly call the built-in data authorization of the platform through the visualization page 、 Data collaboration 、 Identity Authentication 、 Data modeling and other privacy computing applications , Build applications quickly ; Application development users can pass the service level API Interface calls the built-in data privacy protection service of the platform 、 Data privacy computing services and other privacy computing capabilities , Customize your own private computing application ; Technology R & D users can pass the kernel level API Interface calls include homomorphic encryption 、 Homomorphic decryption 、 Property decryption 、 National secret algorithm SM2/SM3/SM4、ID Homomorphic encryption capability including de identification , For multi-party secure computing 、 Technologies such as federated learning converge to form a privacy computing solution .

Homomorphic privacy computing application integration platform has little impact on user business system transformation , Reduce the cost of integrating government users with their business systems . Based on homomorphic encryption technology , The privacy computing application integration platform can keep the original computing power of data after encryption , It does not affect the data flow in the original business system 、 Data usage , The data source only needs to add a privacy computing all-in-one machine to the original system , Data logic 、 Business logic 、 System logic 、 Neither code logic nor communication logic need to be modified , System intrusion is minimized . For example, in the anti fraud system of the public security organ , The public security organ only needs to deploy a homomorphic encryption application system , Provide data protected by homomorphic encryption , Data support for different anti fraud scenarios can be completed .

Homomorphic encryption algorithm developed by homomorphic technology can be used for data exchange among multiple participants 、 Ensure computing performance in scenarios such as data query . One side , The super high speed homomorphic encryption algorithm developed by homomorphic technology is an efficient homomorphic encryption algorithm , It solves the problem of low efficiency of classical homomorphic encryption technology 、 The length of the ciphertext is large , It can significantly reduce the expansion rate of ciphertext , Improve the efficiency of encryption and decryption and ciphertext application , Compare open source homomorphic encryption algorithm , The overall performance is improved by a thousand times . On the other hand , Because homomorphic encryption algorithm encrypts in data layer , No need to build multiple application systems , Relative ratio multiparty secure computing , Homomorphic encryption can greatly reduce the frequency of communication interaction , Save computing power , Improve computing performance .

In terms of security , One side , The super high speed homomorphic encryption algorithm of homomorphic technology is self controllable , And the core modules of relevant software and hardware products are all localized , The core technology is safe and controllable , It can fully realize data security and application security . On the other hand , The technical ability in the financial field has been certified by the testing center under the people's Bank of China 、 The technical ability in the field of password has been certified by the national password detection center 、 The data exchange and sharing mode has been recognized by the National Information Center .

Typical customers ：

The Third Research Institute of the Ministry of public security , Wave cloud , Shanghai data exchange

3.3 Medical privacy computing solutions

Definition ：

The medical privacy computing solution refers to the data flow scenarios in the medical field , Realize invisible privacy computing products and services for user data , It is mainly used in clinical diagnosis 、 Medical research 、 Medical insurance management 、 Pharmaceutical research and development 、 Gene analysis 、 Disease control management and other scenarios .

End user ：

All departments of the hospital , Local health committees 、 The health insurance bureau 、 Big data departments of medical institutions such as the Centers for Disease Control and prevention 、 Technology R & D department , Pharmaceutical R & D departments of pharmaceutical enterprises

Core requirements ：

With the continuous promotion of medical informatization , Medical institutions have accumulated a large amount of medical data , Through artificial intelligence and statistical analysis , These data can be applied to clinical diagnosis 、 Medical research and many other scenes . However , The application of medical data usually needs to collect data from different regions 、 Sample data of different populations , And need to include clinical 、 test 、 Genes and other dimensions . Because the sample data accumulated by a single medical institution is usually limited , Therefore, medical institutions need to use privacy computing technology to jointly model or jointly calculate different data sources . The core requirements of medical institutions for privacy computing solutions include ：

Privacy computing can be applied in a variety of scenarios with different security assumptions and computational complexity . Medical privacy computing application scenarios are diverse and complex , Such as disease diagnosis 、 Medical imaging 、 Genomics 、 Drug target discovery, etc , Various scenarios are important for accuracy 、 The requirements for performance and safety vary greatly , Therefore, medical institutions need to apply a variety of privacy computing technologies , And integrate different technical solutions according to the needs .
The application of a variety of medical professional AI Models and statistical analysis methods to deal with all kinds of medical data . The types of medical data are complex , Contains a variety of structured and unstructured data , Especially genes 、 The complexity of medical images and other data is very high ; meanwhile , Data processing in the medical field is highly specialized , It needs to be widely integrated with medicine 、 biology 、 Professional knowledge in pharmacy and other fields . therefore , When processing data, medical institutions need to apply a variety of AI Models and statistical analysis methods .
Ensure good computing performance in multi-party application scenarios . Medical privacy computing application scenarios , For example, the case comparison with multiple institutions , Study on drug effectiveness , It is often necessary to unite more than ten or more participants . Therefore, medical institutions need to ensure high computing performance in such scenarios .
High computing accuracy in specific application scenarios . Some medical privacy computing application scenarios , Such as patient medication and treatment planning , The calculation results are required to be very accurate , therefore , Medical institutions need privacy computing solutions with high model accuracy in such scenarios .
Link multiple medical data resources in the same region or field . Medical privacy computing applications are often limited by the sample data size , Such as disease diagnosis , The sample size of cases in a single hospital is often limited , Therefore, medical institutions usually need to link multiple data sources in the same region or field to meet the requirements of model training or joint computing .
Meet safety and compliance requirements . Medical data involves a lot of patient privacy , The leakage of these data will cause legal and moral risks , Therefore, medical institutions need privacy computing solutions in data security protection 、 The system environment has high security .

Capability requirements of manufacturers ：

Multi party secure computing 、 Federal learning 、 Homomorphic encryption 、 Differential privacy 、 Trusted execution environment and other privacy computing technologies . Manufacturers need to provide a variety of ciphertext computing technologies , To realize the encryption of different links of joint modeling or joint computing , While meeting the safety requirements , Both performance and accuracy ; For users in the medical industry , In particular, manufacturers need to provide solutions based on trusted execution environment , To reduce the computational complexity , And performance 、 In the scene with high accuracy and security requirements , Meet the needs of users .
Providing professional medical services AI Models and statistical analysis methods . For example, for clinical diagnosis and treatment , Provide auxiliary diagnosis 、 The treatment plan is recommended 、 Medication recommendations, etc AI Model ; For medical image analysis , Provide focus identification, etc AI Model ; For genetic analysis , Provide gene alignment 、 Genome wide association analysis 、 Population stratification and other analysis methods .
Improve computing performance in multi-party application scenarios . One side , Vendors need to be concerned about computational efficiency 、 data compression 、 Bandwidth, etc , Improve the communication efficiency among multiple participants , To improve computing performance . On the other hand , Manufacturers can also schedule third-party computing resources for users through the computing network to improve computing performance .
Providing high-precision federated learning models in specific scenarios . Manufacturers need to improve model accuracy in scenarios that require high accuracy of calculation results , For example, through self-developed federal learning framework , Reduce the precision loss caused by the splitting of the federated learning model .
Establish a data ecosystem in a specific region or domain , Provide users with rich data resources through data operation . The key to the application of medical privacy computing is data resources , Manufacturers need to be in specific areas , For example, at the provincial and municipal level , Or specific areas , For example, research on a certain disease , The joint health commission 、 The hospital 、 The health insurance bureau 、 Pharmaceutical companies and other parties have established a rich data ecology , And provide data operation services .
Privacy computing solutions have high security . Manufacturers need to provide high-level data encryption technology 、 Improve the security design of the platform system to improve the security of the solution . And obtain the safety standard test of the authoritative evaluation organization .

Included in the standard ：

1. Meet the manufacturer's capability requirements for medical privacy computing solutions ;

2. The number of customers served in this market in the past year 3 Above home ;

3. In the past year, the scale of relevant service revenue in this market is 200 More than ten thousand yuan .

On behalf of the manufacturer assessment ：

（ notes ： The evaluation of the following representative manufacturers is sorted according to the phonetic order of the first word of the manufacturer's abbreviation ）

Weiwei Technology

About the manufacturer ：

Product and service introduction ：

For users in the medical industry , Weiwei technology provides Weixin medical big data protection computing platform （NovaVita）. As a sub platform of Weixin privacy protection computing platform , It can meet the needs of users in the medical industry 、 Analysis function 、 Use the differentiated requirements of interaction .

Vendor evaluation ：

Weiwei technology's privacy computing products and services are facing users in the medical industry , Complete platform functions 、 Good experience 、 It can guarantee the good computing performance of multiple participants 、 Provide professional analysis functions for medical scenarios 、 Meet the advantages of high precision and high security .

The privacy protection computing platform of Weiwei technology can adapt to the perfect functions for the needs of users in the medical industry , And provide a good use experience . Weixin healthcare big data protection computing platform (NovaVita) It is a privacy computing sub platform for medical scenarios , The platform is based on the common privacy protection computing platform base of Weiwei technology , Deployable function module groups on the upper layer , Provide matching privacy computing solutions for different customers in the medical field . The platform is helping pharmaceutical enterprises 、 Medical institutions provide safe and efficient data fusion , It also provides medical big data supervision support services , The platform shares case data , combination AI Learning models and other technologies , Empowering medical diagnosis 、 New drug development 、 Specific application scenarios such as gene analysis .

The privacy protection computing platform of Weiwei technology can ensure that large-scale participants （ such as ,10~100 Fang ） Good computing performance in medical scenarios . Weiwei technology is in the design of platform infrastructure , The multi participant joint computing scenario is deeply optimized , While ensuring computing performance , It also solves the problem of communication performance degradation caused by large-scale multi-party data . For example, in the process of convergence model , Through fusion SIMD Technology improves overall computing efficiency ; During communication transmission , Through distributed compression, etc , Optimize the communication efficiency ; In privacy seeking , Optimize bandwidth through symmetric encryption solution .

The privacy protection computing platform of Weiwei technology can provide analysis functions for various professional and complex medical scenarios . Weiwei technology is based on the methodology deposited in the medical field for many years , Select the feature 、 Statistical matching 、 Gene alignment 、 Genome wide association analysis 、 Medical image segmentation and classification and other analysis methods in the medical field are integrated with privacy computing technology to form functional modules , It can realize the medical image 、 Calculation of unstructured data such as genetic data .

The privacy protection computing platform of Weiwei technology can meet the high-precision requirements of some specific medical scenarios on the model . The federal learning framework developed by Kawai technology supports complex lossless model splitting , It effectively makes up for the usual Model Averaging The loss of accuracy caused by , It can ensure that the accuracy of the split federated learning model is numerically equivalent to the calculation accuracy in plaintext after data merging .

Typical customers ：

West China Hospital 、 Xinhua Hospital 、 China Medical 、 Singapore Institute of genetics 、 American Reddy children's hospital

Wing health

About the manufacturer ：

Wing health （Basebit.ai） yes “ Data and computing Internet ” The forerunner of , Is a privacy focused security Computing 、 High tech companies with artificial intelligence and big data . The number of wing square keys is the core of privacy and security computing , Medical service 、 government affairs 、 Finance 、 insurance 、 marketing 、 Science and other fields , Build a data open ecosystem and a data sharing and collaboration environment based on data security and personal privacy protection , And develop the ability of artificial intelligence on this basis , Empower the industry .

Product and service introduction ：

The feature of wing square key number technology is aimed at the whole process of data cross domain circulation , Provide a full stack of technical solutions . Through the privacy and security computing platform, yishufang XDP, Ability to realize data sharing and value acquisition , The core modules include distributed file systems XFS、 Computing resource scheduling and adaptation engine XEE、 Efficient data discovery and integration module DaaS And the safe computing technology path reserved for different trust assumption scenarios PCT, Use full stack technology products to ensure data privacy and security , Unlock data value . Wing number square XDP The platform has data lifecycle management 、 Solid privacy security computing technology system 、 Data driven differentiation AI Apply the three core competencies .

In the field of medical health , Combine privacy security computing with artificial intelligence , For medical institutions 、 Scientific research institutions 、 Biomedical enterprises provide end-to-end full chain data solutions .

For users in the medical industry who have data cleaning and management , Yifang Jianshu also provides an efficient data governance tool based on Artificial Intelligence DataWand.

Vendor evaluation ：

Wing square key is designed to focus on privacy computing , Build a data and computing Internet （IoDC,Internet of Data and Computing）, Provide full stack technology solutions that run through the whole value chain of data , And a wealth of landing cases in the medical industry 、 Self developed privacy computing platform 、 Multi industry and multi Scenario Oriented AI Ability 、 adopt IoDC Provide data and computing resources 、 Platform security and other aspects have advantages , Leading the private computing healthcare industry .

Combine privacy computing with artificial intelligence , Yifang Jianshu has built an end-to-end capability from medical data to intelligent applications , And serve many medical scenarios . For example, a series of intelligent applications based on the medical brain with smart medical records as the core 、 Risk control solutions for medical insurance or commercial insurance 、 Data driven AI Pharmacy 、 Real world research 、 Multimodality omics data analysis collaboration 、 Hospital data asset management platform 、 Infectious disease prevention and control and other medical digital industrialization solutions around data . The goal of privacy computing is to unlock the value of data for customers , In the field of medical health , Especially for all kinds of medical institutions at all levels , Need to have full stack technology in the medical field, including from data collection to data governance , From privacy computing to intelligent tool model construction , Finally, various intelligent applications will be formed to provide landing services in different scenarios . Wing - to - wing keys build this end - to - end capability , And realize the construction of a data-driven future smart hospital in the head class III hospitals such as Ruijin Hospital .

In the wing number square XDP In the platform , Secure data storage 、 Convergence and discovery 、 Computing resource scheduling and privacy computing technology , Be able to extract value from data , Realize the management and circulation of the whole life cycle of data elements . Built in a variety of self-developed privacy computing technologies , It can meet the user requirements of different security trust assumptions and computing complexity scenarios . Safe sandbox XSSBox It can provide a local computing environment with zero trust for a single platform ; Trusted execution environment XTEE for XDP The computing tasks within and between platforms provide a hardware based trusted execution environment ; Self developed federal learning framework XFL Support includes horizontal 、 Vertical and federated transfer learning and other mainstream algorithms , performance 、 The accuracy and security of the model are better than the mainstream open source framework ; Self developed ciphertext computing framework SXCE Can integrate MPC/FHE/DP/ZKP And other privacy computing technologies , It can be used for ciphertext calculation with different accuracy and performance requirements , Can also cooperate with federal learning , Enhance the security of joint modeling .

Yifang Jianshu is suitable for medical scenarios AI Model and medical knowledge base , Able to process complex multimodal medical data . Medical knowledge base is based on machine learning 、 Knowledge map technology constructs 200 ten thousand + Medical concepts 、2000 ten thousand + Medical relations 、10 ten thousand + Medical reasoning rules 、500 ten thousand + Mutual mapping medical coding . The platform has built-in a large number of self-developed in the medical field AI Model , For example, after the medical record is structured 、 The source of infection spreads 、 Auxiliary diagnosis 、 Supplementary Examination / Medication 、 Image focus recognition 、 Medical insurance fraud monitoring 、DRG/DIP Intelligent fee control, etc , Realize data-driven empowerment for various medical scenarios .

Besides , Based on wing number workshop XDP Built IoDC It can help users reach more data and computing resources . The number of wing keys is calculated through privacy “ Data available invisible ” The ability of , On the premise of protecting data privacy and security , By building a data network , Help data in manageable 、 Measurable and protected by privacy security, it provides data assistance for artificial intelligence . Built on the wing IoDC in , Through the privacy computing power of the platform , And distributed file systems 、 Data discovery and integration 、 Computing resource scheduling and adaptation , Each node has complete data value realization capability . One side , The user can go through IoDC Use authorization data from various government and medical institutions . for example , In Yichang City, Hubei Province , Through opening up multiple internal and external data sources of government and enterprises and cross platform joint Computing , Carry out multi-point trigger early warning of infectious diseases . Through government enterprise data fusion , Data collaboration of all parties , Use different data sources to screen out high-risk groups of infectious diseases , Precise screening 、 Eliminate false positives , Realize data-driven intelligent infectious disease prevention and control . On the other hand , adopt IoDC Deploy the Supercomputing Center 、 Data and computing resources in Colleges and universities . For example, Yifang Jianshu cooperates with Nanjing Jiangbei new area biomedical public service platform , Call data and computing resources , In the propulsion area IoDC, It can solve the scientific research of biomedical industry in a one-stop way 、 Production and clinical transformation 、 Practical problems such as drug research and development .

In terms of security , Yifang health data can fully meet the needs of medical users for data security and compliance . Yifang Jianshu participated in the formulation of China Communications Standardization Association 、 A number of privacy computing security projects organized by China Academy of information technology 、 Interconnection standards , Wing number square XDP The platform has passed the ICT Institute “ Trusted execution environment ”、“ Federal learning ”、“ Multiparty secure computing ” Relevant ability evaluation and credible selection evaluation of health care big data .

Typical customers ：

Yichang municipal health commission 、 Xiamen health care big data management center 、 Ruijin Hospital Affiliated to medical school of Shanghai Jiaotong University 、 West China Second Hospital of Sichuan University 、 Cancer prevention and treatment center of Sun Yat sen University

3.4 Privacy computing platform

Definition

Privacy computing platform , It refers to secure multiparty computing 、 Federated learning and other privacy computing technologies , On the premise of providing privacy protection for data , Invisible platform tools are available for data circulation . The privacy computing platform has general privacy computing technology service capabilities , It can support enterprise users to build various privacy computing application solutions .

End user ：

Finance 、 government affairs 、 Medical care 、 retail 、 telecom 、 Enterprises or institutions in various fields such as transportation IT department 、 Big data Department 、 Scientific and technological innovation department

Core requirements ：

As data becomes an innovative business of enterprises in various industries 、 Key elements of optimizing operation management , Cross organizational data circulation and sharing has become an increasingly important demand of enterprises . meanwhile ,《 Network security law 》、《 Data security law 》、《 Personal information protection law 》 And other laws and regulations have been introduced , Many compliance requirements have been added to the data flow . therefore , In recent years, many enterprises and institutions in various industries have begun to pay attention to privacy computing , And hope to build a private computing platform , Build infrastructure for further exploring the application scenarios of privacy computing . The core requirements of enterprises for privacy computing platforms usually include ：

It has high versatility . The potential application scenarios of enterprises are very wide , However, at the initial stage of platform deployment, these application scenarios are often not clear . therefore , Enterprises need the platform to have high universality , It can meet the potential application scenarios and functional requirements of the enterprise in the future .
Flexible operation mode . The customization degree of different privacy computing application scenarios varies , Combine computing processes from calling operators , To the custom calculation process , Its requirements for users' technical level are also different . therefore , Enterprises need the platform to have a variety of operation modes , Flexible to meet user needs .
Easy to deploy and integrate . One side , Enterprises hope that privacy computing applications can be quickly implemented and produce results , Therefore, it is required that the solution can be deployed in a convenient and fast way ; On the other hand , Many enterprises have established more complex business and IT System , Therefore, the privacy computing solution is required to minimize the transformation of the original system , Integration with the original system .
High performance . In joint modeling 、 In offline scenarios such as joint statistical analysis , With the increase of data size , And the enrichment of application scenarios , Its computing performance will be restricted ; And in joint forecasting 、 In scenarios with high real-time requirements such as stealth query , As the number of requests increases , The calculation delay will gradually appear . Enterprises need the platform to have high performance in these scenarios .
With high security . To ensure the security of data assets , And to meet the requirements of relevant laws , Enterprises need a platform for data security protection 、 System environment 、 The interpretability of the computing process meets the high security requirements .

Capability requirements of manufacturers ：

The platform is universal in terms of privacy computing technology and system functions . In terms of privacy computing technology , The platform needs to have federal learning 、 Multi party secure computing and other privacy computing technologies , Can support joint modeling 、 Joint statistics 、 Privacy seeking 、 Various application scenarios such as stealth query ; In terms of system functions , The platform needs to modularize the system functions , Support users to add function modules according to their needs , Such as blockchain deposit certificate 、AI Calculation 、SQL Equal module .
The platform has flexible operation and use modes . One side , The platform needs to integrate multi-party secure computing 、 Algorithms such as federated learning are split into refined operators , Support users to graphically combine operators according to their needs , Build the computing process ; On the other hand , The platform also needs to support some users to adopt in scenarios with a higher degree of customization Python and SQL And so on , Customize the calculation process .
The platform can be rapidly deployed and integrated . In terms of deployment , In addition to the local deployment mode , The platform needs to provide agile deployment and delivery methods , For example, the platform adopts cloud native architecture , And support containerized delivery ; With SDK or API Support users to quickly build privacy computing applications ; In terms of integration with the original system , The platform needs to provide componentized and interfaced services to support the data exchange between the privacy computing platform and the original system 、 account number 、 Log and other aspects of rapid docking .
The platform has high performance in various offline and real-time scenarios . In an offline scenario , The platform needs to support large-scale distributed computing and hardware acceleration , Improve computing performance ; In a real-time scene , The platform needs to be deeply optimized in terms of communication efficiency , And ensure the stability of real-time calculation , Reduce the delay of real-time computing .
The platform has high security . Manufacturers need to provide perfect data encryption technology 、 Improve the authority control of the platform to improve the security of the platform ; And need to support algorithm flow visualization , And support access to third-party traffic audit tools to verify the use of data, etc. to improve the interpretability and reliability of the computing process .

Included in the standard ：

1. Meet the manufacturer's capability requirements of the privacy computing platform ;

2. The number of customers served in this market in the past year 5 Above home ;

3. In the past year, the scale of relevant service revenue in this market is 500 More than ten thousand yuan

4、 List of selected manufacturers

原网站

版权声明
本文为[Love analysis ifenxi]所创，转载请带上原文链接，感谢
https://yzsam.com/2022/162/202206110343528714.html

当前位置：网站首页>2022 love analysis · privacy computing vendor panoramic report | love Analysis Report

2022 love analysis · privacy computing vendor panoramic report | love Analysis Report

边栏推荐

猜你喜欢

随机推荐