FedAI.org https://www.fedai.org Federated AI Ecosystem Tue, 28 Apr 2020 11:26:15 +0000 en-US hourly 1 https://wordpress.org/?v=5.3.1 WeBank Tops the 2019 Global Banking Patents Rankings https://www.fedai.org/news/webank-tops-the-2019-global-banking-patents-rankings/ Tue, 28 Apr 2020 11:04:58 +0000 https://www.fedai.org/?p=1847


In the recently published 2019 Global Banking Patents Rankings (Top 100) by IPRdaily, WeBank, a world leading digital bank, has jumped from the fifth place (2018 Rankings) to top this year’s rankings with 632 filed patents.

Among all patents WeBank filed in 2019, 80% of them related to ABCD technologies (AI, Blockchain, Cloud Computing, Big Data).

In terms of AI, WeBank constructs a technology-driven financial ecosystem with advanced technologies of federated learning, new generation of human-computer interaction, AI marketing and AI asset management. Its AI patents cover federated learning, NLP/Intelligent Speech, machine learning, advertising/recommendation tech, computer vision, and more. When it comes to blockchain, WeBank has a complete set of open source consortium chain infrastructure, middleware and solutions with patents of algorithm, privacy protection, cross-chain tech, etc.. According to a recent report from the Block, WeBank filed the third most blockchain patents globally in 2019. It is also the only bank on the top 10.

An increasing number of patents filed indicates the banking industry’s growing fintech capabilities. Rather than build a heavily guarded fortress with those capabilities, WeBank determines to empower developers with open source fintech in a collaborative ecosystem. In 2019, WeBank has announced the open source fintech strategy that targets to collaborate with developers around the world. So far, it has brought out dozens of successful open source projects, such as FATE(Federated AI Technology Enabler), FISCO BCOS (consortium chain platform built together with FISCO open source taskforce team) and open-source big data platform suite WeDataSphere.

IJCAI 2020 Tutorial https://www.fedai.org/research/conferences/ijcai-2020-tutorial/ Wed, 11 Mar 2020 07:22:32 +0000 https://www.fedai.org/?p=1828



Federated Recommender System


The recommender system (RecSys) plays an important role in the real-world applications, from product recommendations to news recommendations. It has become an indispensable tool for coping with information overload. In general, RecSys is heavily data driven, the more data is used in the Recsys, the better recommendation performance can be obtained. They collect the private user data, such as the behavioral information, contextual information, the item metadata, the purchase history, the recommendation feedback, the social information and so on from different data sources. All this informative user data is centrally stored at the database of each organization to support different kinds of recommendation services.

However, collecting data from multiple parties could lead to serious privacy and security risks and violate laws such as the General Data Protection Regulation (GDPR). In this tutorial, we introduce a new notion of Federated Recommender System (FedRec). Compared to the conventional RecSys, FedRec protects the user privacy and data security through decentralizing private user data locally on each party. In this tutorial, first of all, according to the data structure of recommendation tasks, we categorize the FedRec into horizontal FedRec Systems, vertical FedRec systems and transfer FedRec systems. We explain the problems and typical scenarios in each category. Secondly, we implement an open-source tool that contains typical algorithms in each category. The tool is based on FATE, which is an industrial level framework designed to support federated learning architectures and secure computation. We demonstrate the idea and implantation details of each FedRec algorithm. Thirdly, we show two applications in news recommendation and online advertising. Finally, as a promising direction with huge potential opportunities, we discuss challenges and open questions in FedRec systems.



Part I:  Theory and Algorithms of FedRec Systems (~1:45hour)

1) Preliminary of Recommendation (~20min)

2) Preliminary of Federated Learning (~30mins)

3) Categorization and typical algorithms (~50mins)

Part II:  Implementation and Applications of FedRec Systems (~1:45hour)

4) FATE: open-source project to support the federated AI ecosystem (~20mins)

5) FedRec algorithms and their implementations (~50 mins)

6) Applications and future works (~35 mins)


Slides in PDF(Upcoming stay tuned)

Open-sourced Project



·Qiang Yang: Chair Professor at Department of Computer Science and Engineering, Hong Kong University of Science and Technology; Chief AI Officer, WeBank, China

Prof. Yang is currently a Chair Professor in the Department of Computer Science and Engineering of the Hong Kong University of Science and Technology, and Chief AI Officer at WeBank, China. He received his MSc degree in Astrophysics, MSc in Computer Science and PhD in Computer Science at University of Maryland, USA.

His research interests include federated learning, transfer learning, machine learning, planning and data mining in artificial intelligence. He is also the President of IJCAI, Executive Council Member of AAAI, and Editor in Chief of IEEE Transactions on Big Data. He was the founding Editor in Chief of ACM Transactions on Intelligent Systems and Technology. He was the founding director of the HKUST’s Big Data Institute, the founding director of the Huawei Noah Ark Research Lab, and the founding director of Wechat-HKUST Joint Lab on AI. He is a fellow of ACM, IEEE, AAAI, AAAS, IAPR and CAAI.

·Ben Tan: Senior researcher in the AI department of WeBank, China

Dr. Tan is currently a senior researcher in the AI department of WeBank, China. Previously, he was a senior researcher at Recommendation Center, Tencent, China. He received his PhD in Computer Science at the Hong Kong University of Science and Technology.

His research interests include Recommender systems, online advertising, federated learning, transfer learning. He has over five years experiences on recommender systems and online advertising. He has published over ten papers on several top-tier conferences and journals such as KDD, AAAI, SDM, Pattern Recognition, IEEE Transactions on ITS, ACM Transactions on IST, and presented his work in multiple external venues. He serves on the program committee member of WWW, IJCAI, CIKM, SDM and related conferences.


Contact Us:

Email: ai.ads@webank.com

Official Stie: https://ad.webank.com/

The WeBank AI Group Present the First Monograph on Federated Learning https://www.fedai.org/news/the-webank-ai-group-present-the-first-monograph-on-federated-learning1815/ Mon, 09 Mar 2020 07:16:21 +0000 https://www.fedai.org/?p=1815


In the big data and AI era, how to effectively utilize the decentralized data? How to address data security and enabling privacy? The first monograph on federated learning- <Federated Learning> gives you answers.

<Federated learning> is from the renowned AI-series published by Morgan & Claypool Publishing House, written by six leading AI experts in two years. This monograph shares the experience and insights of the WeBank AI group in the field of federated learning in promoting privacy-preserving AI and collaborations among different organizations.


What does the monograph tell us?

In this book, we describe how federated machine learning addresses this problem with novel solutions combining distributed machine learning, cryptography and security, and incentive mechanism design based on economic principles and game theory. Federated learning provides a feasible solution that the data owners can share the machine learning models while ensuring that no local data leaves any data owners. This book presents recent advances in federated learning research, open-source platforms and potential applications, as well as examples of practical use cases of federated learning in finance and computer vision, etc. This monograph shows that federated learning is a fundamental building block of the next generation AI. It enables large-scale collaboration among geo-distributed data owners.

The content of this book:

  • The overview of federated learning: problems to be solved, definition, classification, development, etc.;
  • Background knowledge of federated learning, including the privacy protection of machine learning technology and data analysis;
  • Introduction of distributed machine learning. Emphasizing the difference between federated learning and distributed machine;
  • The definition, framework, algorithm of horizontal federated learning, vertical federated learning, and federated transfer learning
  • Incentive mechanism design based on economic principles and game theory
  • The practical use cases of federated learning in computer vision and Natural language processing and recommendation systems
  • Federated reinforcement learning
  • The potential applications of federated learning in different industries, such as finance, medical treatment, education, smart city


What can you learn from this book?

We are in an era with massive but fragmented data. We need a solution to utilize the scattered data and produce added-values, while ensuring user privacy and data security. Federated learning provides a feasible solution for building AI using decentralized data and for protecting user privacy and data security. The practice of federated learning is the new dawn of AI.

This book provides with the readers a new perspective about building privacy-preserving AI, especially for those who are concerned with data abuse and user privacy in the big data and AI industry.

Please click here for purchasing the English version! The Chinese version will be available in April. Coming soon!


About WeBank AI Group

This monograph is written by WeBank AI group leading by the chief artificial intelligence officer (CAIO) Qiang Yang. Prof. Qiang Yang is the pioneer of transfer learning and federated learning. He is the first Chinese scholar who became a fellow of the Association for the Advance of Artificial Intelligence (AAAI) and a member of the AAAI executive committee. Prof. Qiang Yang is the chairman of the council of the International Joint Conference on Artificial Intelligence (IJCAI), the first Chinese scholar to serve as the chairman. He won the 9TH WU WEN JUN AI OUTSTANDING CONTRIBUTION AWARD in 2019, which is a very prestigious award in the area of AI in China.

The WeBank AI group is a top AI research team of WeBank. It aims to use the autonomous and controllable, safe and dependable AI technology to explore the new way of FinTech, leading the new direction of the AI industry. We have achieved great successes in the following four fields: The FedAI ecosystem, new generation human-computer interaction, precision marketing, and smart asset management.

In the field of federated learning, the WeBank AI Group have initiated the establishment of IEEE standard and domestic standard on federated learning. The WeBank AI Group have also released the world’s first industrial-grade open-source federated learning platform, known as FATE (Federated AI Technology Enabler). We keep exploring possible business opportunities and applications of federated learning, and seeking to solve practical problems. We endeavor to promote the FedAI ecosystem and welcome collaborations from both academic and industrial partners. We try to build trustworthy and responsible AI under data protection laws and regulations.

WeBank AI group cooperates with Mei Cai to help upgrade the fresh-food industry https://www.fedai.org/news/webank-ai-group-cooperates-with-mei-cai-to-help-upgrade-the-fresh-food-industry/ Tue, 11 Feb 2020 10:29:15 +0000 https://www.fedai.org/?p=1807


Introduction: With the improvement of people’s living and the policy support, the fresh food industry develops rapidly. The non-standard problem of logistics and storage will become more and more important. Helping the efficiency upgrade in the fresh food industry, the cooperation between WeBank and Mei Cai is using the AI technology to manage warehouse and conduct the sales forecasting.

In recent years, with the development of new retail, the demand for the fresh food industry has exploded, showing great potential development. Meanwhile, under the policy of supply-side structural reform, the relevant departments will strengthen agricultural product circulation facilities and market construction and thus the quality of agricultural products is improved obviously. According to the reference, the volume of business in China’s fresh food market is increased steadily from 2011 to 2018. The volume of business reaches 1910 billion yuan, with year-on-year growth of 6.7%. It is estimated that the volume of business in China fresh food market will reach 231 billion yuan by 2020, with a compound annual growth rate of 14.16%.

Although the fresh food industry develops quickly, it also faces a lot of problems. The fresh-food product has a short shelf life, the difficulty of cold chain distribution and non-standard features. The warehouse storage of fresh food, the consumption of products and the distribution will be the challenges for the practitioners. Recently, the WeBank AI group cooperates with Mei Cai company to help improve the smart warehousing and sales forecasting and optimize the efficiency by the AI application.

Focusing on the issues of high commodity wastage rate (high percentage of damaged goods) and the difficulty in control of preparing goods, the two parties will develop the smart warehousing by analyzing the business district, high-speed intersection, the location of the wholesale market, subsidy policy and other dimensions. Through setting a distribution efficiency and performance conditions, the smart warehousing can evaluate the traffic, the surrounding situation and the environmental quality inside the warehouse, to assist in setting up a warehouse and moving a warehouse, and other decisions.

As for the sales forecasting, according to the numerous marketing segmentation scenarios of fresh food e-commerce, WeBank and Mei Cai develop forecasting technologies such as daily sales forecasting, multi-day sales forecasting, real-time sales forecasting. Every forecasting technologies will combine the advantages of multiple algorithm models. They not only combine the ARIMA, FBprophet and other time series algorithms, including Random Forest, XGBoost, and other machine learning algorithms, but they also develop some particular model which is suitable for Mei Cai business. In some specific business scenarios, Mei Cai will adopt business-fit strategies based on the prediction algorithm. For example, in the unsold vegetable program, Mei Cai reduces the percentage of the unsold vegetables on the day to 6.9% and the percentage of vegetables purchased in the morning is limited to 12.5%.

At the same time, they will keep developing the business logic of the fresh food e-commerce and modeling the business logic. They establish lots of basic characteristic models such as promotion model, price elasticity model, sales attribution analysis, commodity substitution relationship model, business profile, and business consumption-ability model. Mei Cai will apply the modeling result to the sales forecasting for the further improvement of forecasting accuracy rate. In the meantime, it will produce a business optimization strategy.

In the future, WeBank and Mei Cai will continue to drive the technology revolution of the fresh food supply chain industry and establish a promotion platform which can keep optimizing the sales forecasting, based on the automated machine learning. Then they will begin to research and develop the AI smart pricing and intelligent contact system. If you want to learn more cases, please go to our website: www.fedai.org

About WeBank AI group:

WeBank AI group is an artificial intelligence research team under Qianhai WeBank Bank Co., Ltd (WeBank). It aims to use the autonomous and controllable, safe and dependable AI technology to explore the new way of Fintech, leading the new direction of the AI industry. Moreover, we are based on industrial application scenarios, starting from finance. We also have a lot of explorations in the frontier of AI with representative achievements including federated learning, transfer learning and so on. As a leader of federated learning, WeBank not only proposes the new direction of “federated transfer learning”, but also leads and promotes the ecosystem construction of AI cooperation under the protection of data privacy over the world.

About Mei Cai:

Mei Cai was founded on June 6th of 2014. Mei Cai always tries to use the internet thinking to change China’s agricultural market. By its particular mode: one chain and one platform for farmers and restaurants, Mei Cai upgrade the agricultural product supply chain to improve the circulation efficiency for benefiting both sides. They focus on providing a full category of one-step food procurement services for the national nearly 10 million restaurants. Mei Cai keeps the fast growth in recent 5 years.

IBM and China’s Digital Bank WeBank Jointly Held Workshop on Federated Learning https://www.fedai.org/news/ibm-and-chinas-digital-bank-webank-jointly-held-workshop-on-federated-learning/ Mon, 10 Feb 2020 03:01:43 +0000 https://www.fedai.org/?p=1800


NEWYORK, USA, Feb. 6, 2020. IBM and China’s leading digital bank WeBank jointly organized “Workshop on Federated Learning and Analytics (FL-IBM’20)” at IBM T.J. Watson Research Center. Experts and scholars from IBM, WeBank, Google, MIT, University of Minnesota and other institutions participated in the meeting. New methods of federated learning were shared by 9 invited speakers, and ideas were exchanged at the panel discussion on“Data Privacy and Regulatory Issues in AI: Enterprise and Customer Perspectives”.

As a new paradigm of distributed encrypted machine learning, Federated Learning(FL) enables all parties to build models jointly without sharing data, that is, to connect data silos without violating data privacy regulations. In the past two years, more and more researchers and enterprises have paid attention to Federated Learning. How to improve the efficiency and performance of FL? How to develop Incentive mechanism to attract more participants? How can FL be applied at a truly large scale? Current challenges faced by FL and the future direction of FL were discussed at the workshop.


Official website of workshop:https://federated-learning.bitbucket.io/ibm2020/

Know more about Federated Learning:https://www.fedai.org/

WeBank AI Group publishes its first Paper in ACM CHI 2020 https://www.fedai.org/news/webank-ai-group-publishes-its-first-paper-in-acm-chi-2020/ Mon, 10 Feb 2020 02:27:17 +0000 https://www.fedai.org/?p=1793


Recently, ACM SIGCHI Conference on Human Factors in Computing System (ACM CHI 2020 for short) released its paper acceptance result. A collaborative paper with the title of “MaraVis: Representation and Coordinated Intervention of Medical Encounters in Urban Marathon” (MaraVis for short) between WeBank AI Group, Tencent CSIG Jarvis Lab, and HKUST was accepted by CHI 2020, which is also the first paper accepted by the top conference in the field of human-computer interaction from WeBank.

CHI conference is one of the three A-level conferences in the field of human-computer interaction and pervasive computing. The papers accepted by the CHI conference are world-recognized for the high quality. It is reported that CHI 2020 received 3126 papers and the paper acceptance rate is only 24.31%. The acceptance of MaraVis also represents the international recognition of WeBank’s achievements in the field of visual analytics and interactive intelligence.

There is an increased use of Internet-of-Things and wearable sensing devices in the urban marathon to ensure an effective response to unforeseen medical needs. However, the massive amount of real-time, heterogeneous movement and psychological data of runners impose great challenges on prompt medical incident analysis and intervention. Conventional approaches compile such data into one dashboard visualization to facilitate rapid data absorption but fail to support joint decision-making and operations in medical encounters. This paper presents MaraVis, a real-time urban marathon visualization and coordinated intervention system. MaraVis first visually summarizes real-time marathon data to facilitate the detection and exploration of possible anomalous events. Then, it calculates an optimal camera route with an arrangement of shots to guide offline effort to catch these events in time with a smooth view transition.

System Overview of MaraVis

WeBank AI Group also launched visual analytics applications in other fields. For example, WeBank AI Group establishes an intelligence system covering retail goods sales prediction and facility location issues in the field of smart retail. To be specific, the team launched a retail goods sales prediction system which forecasts the sale volume in hour-level. By combining the historical sales data, the insight of customers’ profiles can be obtained to optimize the overall goods planning, which can further update the traditional retail industry mode from “production-based selling” to “selling based production”. Moreover, the team combines the frontier technology of visualization and visual analytics with professional domain knowledge and proposes resolutions to the pain points in the industry, forming the interactive “human-in-the-loop” analysis. The team also proposed visual analytics solutions in the fields of risk management, smart retail and supply chain, medical health, enterprise management, and urban traffic management.

The successful launch of the above applications cannot be separated from the theoretical support. The team continuously strengthens its theoretical research. For example, in July of 2019, WeBank AI Group won the “Visual Analytics Design Award” in the data visualization competition of the 6th China Visualization and Visual Analytics Conference (ChinaVis 2019); The team’s work “Multi-Agent Visualization for Explaining Federated Learning” was also showcased in the top AI conference IJCAI 2019. For more cases, please visit our website: https://vis.webank.com.

The acceptance of MaraVis represents that WeBank has achieved great progress in the field of visual analytics and interactive intelligence. In the future, WeBank will collaborate with more business partners in both academia and industry. WeBank is also committed to exploring a new path of fintech with autonomous, controllable, safe and reliable AI technology and leading a new direction of the AI industry.

To know more about WeBank AI group, please visit https://ai.webank.com

Federated Learning https://www.fedai.org/research/book/federated-learning/ Thu, 06 Feb 2020 13:42:33 +0000 https://www.fedai.org/?p=1783


December 2019


Qiang Yang

Yang Liu

Yong Cheng

Yan Kang

Tianjian Chen

Han Yu


How is it possible to allow multiple data owners to collaboratively train and use a shared prediction model while keeping all the local training data private? Traditional machine learning approaches need to combine all data at one location, typically a data center, which may very well violate the laws on user privacy and data confidentiality. Today, many parts of the world demand that technology companies treat user data carefully according to user-privacy laws. The European Union’s General Data Protection Regulation (GDPR) is a prime example. In this book, we describe how federated machine learning addresses this problem with novel solutions combining distributed machine learning, cryptography and security, and incentive mechanism design based on economic principles and game theory. We explain different types of privacy-preserving machine learning solutions and their technological backgrounds, and highlight some representative practical use cases. We show how federated learning can become the foundation of next-generation machine learning that caters to technological and societal needs for responsible AI development and application.

WeBank and Terminus implemented the first federated trace forecasting model, and established the “AIoT Joint Laboratory” https://www.fedai.org/news/the-first-federated-travel-forecasting-model-is-implemented-in-beijing-webank-and-terminus-jointly-establish-the-aiot-joint-laboratory/ Thu, 16 Jan 2020 10:36:58 +0000 https://www.fedai.org/?p=1777


On the 23rd of December, Webank and Terminus announced officially that they have established the “AIoT Joint Laboratory”. The two parties will jointly research the field of “Artificial Intelligence + Internet of Things” (AI + IoT), combined with WeBank’s AI technology represented by federated learning and the rich business experience of Terminus ‘s urban management. The cooperation will help upgrade the smart city and intelligence security, such as urban management, public security and community. The deputy manager of the WeBank AI group, Tianjian Chen and Yi Zhang, the general manager of the ecosystem product center of Terminus technology group, inaugurated the joint laboratory and signed the cooperation agreement.


From IoT to AIoT, AI becomes the key to the transition

In the recent year, with the rapid development of IoT technology, “smart city”, “intelligent security” and “intelligent transportation” has begun to come into reality. However, during the process of the transition from IoT to AIoT, the development of technology is quite bumpy and facing many challenges:

One is the information silo. Taking the implementation of smart cities as an example, the different department manages its information. It is difficult for them to share information and thus it became the information silo, which increased the difficulties of management and decision-making.

Second is the “non-standard” situation. With the city development, the magnitude of data keeps expanding. Because of the lack of uniform interface standards and the difficulty to collaborate and integrate the specification, the value of data is hard to play effectively.

The third is the need for “data security” on how to search, storage and use the data based on “safety compliance”, avoiding to abuse data. This is the common request of self-privacy protection from the current legal regulation and public. It is also a key technology in the development of the industry.

Under this background, WeBank cooperated with Terminus to establish “AIoT Joint Laboratory”. It will open a new development of AIoT through the federated learning technology to overcome the difficulty of IoT data. As the new generation AI technology, federated learning can jointly build models to improve the performance of AI models under the premise of participant’s data never leaving the local storage. It is reported that the frontier of AI technology, federated learning is pioneered by WeBank in China. Currently, it has been applied to smart loans, smart risk management, anomaly detection and, other business scenarios to help promote the smart city, intelligent finance and other industries. However, as a leader of smart city-class IoT platforms in China, Terminus has served lots of scenarios in the AIoT applications, such as public security, population, fire control, police service and travel, owning a deep industry experience. With the powerful AI ability, WeBank combined the rich business experience of Terminus. They will greatly accelerate the transition from IoT to AIoT, thereby bringing a broader application market for the IoT industry.

WeBank and Terminus focus on AIoT, the result of “smart community” is outstanding

The “AIoT Joint Laboratory” choose the “Smart community” as the starting point to accelerate AIoT research and practical application. The federated trace prediction model researched by the cooperation between WeBank and Terminus has been implemented officially in some areas of Beijing. It is known that, based on the community security data, federated model connects many communities and combines the timing and space of the trip, and historical trip experience under the premise of data keeping in the private storage of the community server. It will use the federated learning technology for modeling with the real-time data monitoring and AI visual analysis method, to intelligently forecast the timing and location that the authorized person who needs help may appear in it. Therefore, it can analyze intelligently different peoples. While providing targeted services for community residents, it also provides technical support for delicacy management of community and urban population.

For example, the federated trace forecasting model can conduct the feature extraction of the track for solitary elderly and build model and judges the behavior of the elderly based on the model. If there is an abnormal deviation at the monitoring time, it is judged as a trajectory anomaly. In this way, different warning rules can be set for the elderly who have not been out for a long time and corresponding warnings can be given to inform the relatives of the elderly and the community management staff, so they can provide help as soon as possible.

The trace forecasting model of the elderly

Besides, the community neighborhood committee and management office can know the life of old mem according to the access records. Combined with the early warning system, the staff can visit the specific person. It not only keeps residents safe but also it reflects the care of people’s livelihood in the community, helping the construction of security and community.

In addition to providing security warnings and targeted care for vulnerable groups such as the elderly and children, this model is also suitable for community security, such as through precautionary control of illegal drug users to change the traditional person-to-case investigation mode to avoid threats and irreversible negative impacts on the lives and property of the people.

The implementation of the federated trace forecasting model not only can provide a more safety and comfortable living environment for the community residences, but it also reduces the management cost of property management and raise the management level of the intelligence, digitization and convenience, accelerating the coming of “smart community”.


Keep accelerating the implementation of AIoT

The federated trace forecasting model is implemented in the “smart community”, which is the good beginning of “AIoT Joint Laboratory”. It is also another milestone in the development of AIoT, but it is far from the end. In the future, based on the cooperation to develop the smart community prediction model, the two parties will keep focusing on the AIoT and improve the federated learning technology, unleash the potential of multidimensional data. In the cooperation of AI technology and big data, more industrial applications and solutions will be generated to support richer scenario requirements and help upgrade more intelligence industry. As the deputy manager of the WeBank AI department, Tianjian Chen said, “Technology must land in the business scene. AI can genuinely liberate the human workforce, thereby enabling the technology inclusive industry, the general public, and the inclusive society.” Only in this way, lots of the “smart community” can connect one point after another, gradually form a line. Then form a network of a smart city for us to have intelligent life.


About WeBank AI group:

WeBank AI group is a top artificial intelligence research team under Shenzhen Qianhai WeBank Co., Ltd. The AI team is committed to exploring new ways of Fintech with autonomous, controllable, secure and reliable AI technology, and leading the new direction of the AI industry. Based on the industry application scenarios and starting from the finance, the AI team has many explorations on the frontier of AI. The representative of achievements included federated learning, transfer learning and so on. As a leader in federated learning, WeBank not only proposes the general solution “federated transfer learning” for the first time in all over the world, but it also leads and promotes the AI collaboration ecosystem construction under the protection of data privacy.


About Terminus:

Terminus is a new technology company under China Everbright Group’s development strategy and also an intelligent IoT unicorn company supported by China Everbright Group. Terminus takes the lead in proposing the AIoT technology structure and uses it. Providing the intelligent technology service of public management and public service for government and enterprises, it aims to become the smart scenario service company leading in the world and create industry-leading solutions in the scenarios of community, public utilities, electricity and energy, and public service. Terminus also set up research centers in Beijing, Shanghai, Chongqing, Wuhan, Shenzhen and other cities. Terminus has owned 720 domestic patents including 417 patents for innovation and it is selected into the authoritative IT research and consulting company Gartner report for six times and has been unanimously recognized by the industry.

More information about federated learning, please come to our FedAI official website: https://www.fedai.org/

NuerIPS 2019: China’s WeBank, Mila, and Tencent Partner on AI Federated Learning to Protect Data Privacy https://www.fedai.org/news/nuerips-2019-chinas-webank-mila-and-tencent-partner-on-ai-federated-learning-to-protect-data-privacy/ Wed, 25 Dec 2019 08:16:08 +0000 https://www.fedai.org/?p=1759


VANCOUVER, Dec. 13, 2019 – Top AI conference NeurIPS 2019 was held in Vancouver from December 8-14th. Attending experts were excited about a new research direction named federated learning (FL). Professor Yoshua Bengio, A.M. Turing Award Winner, founder of the world’s top deep learning research facility Mila-Quebec Artificial Intelligence Institute and one of the “three musketeers of deep learning”, said that “In terms of better training neural networks, federated learning is at the forefront of research and will have important impact on business. ”

Currently, data silos and privacy protection are two big challenges for AI. As an encrypted distributed machine learning framework, FL can tackle both problems by allowing different parties to build models collaboratively without the need to reveal their data. The method helps to advance AI modeling while protecting data and privacy.

China’s digital bank WeBank is a leading research facility in federated learning. At NuerIPS 2019, WeBank co-organized the FL workshop with Google, CMU, and NTU, with 400 scholars joining in the discussion.

During the WeBank AI Night event, WeBank announced two strategic partnerships with Mila and the leading cloud computing platform Tencent Cloud. The cooperation will focus on further developing federated learning, based on WeBank’s real-world experiences in finance and fintech, adhering to Mila’s core philosophy “AI for Humanity”, Tencent’s “AI for Good” and WeBank’s “Make Banking Better for All ” to create safe, inclusive AI applications.

Professor Qiang Yang, WeBank’s chief AI officer, explained that large-scale AI application relies on big data, which is scattered across many different organizations. Direct data merging will violate privacy regulations. FL is a compliance method strictly following laws and regulations, and is now used in fintech, healthcare, smart city, and other industrial applications.

To reduce the use threshold of federated learning, WeBank launched the world’s first industry FL open-source framework Federated AI Technology Enabler (FATE) in February 2019. This grants a ready-to-use FL framework tool to any companies wishing to work together. Partner Tencent Cloud and companies including Huawei, JD.com and other tech giants have all joined the ecosystem. The company is also leading international IEEE standards on the technology.

Founded in 2014, WeBank is the world’s leading digital bank operating solely online, now serving over 170 million individual customers and over 500,000 small and micro-sized enterprises.

The first Semi-Interactive demo of federated learning appears in the Shanghai art exhibition. Let’s experience the AI game charm with zero distance https://www.fedai.org/news/the-first-semi-interactive-demo-of-federated-learning-appears-in-the-shanghai-art-exhibition-lets-experience-the-ai-game-charm-with-zero-distance/ Wed, 18 Dec 2019 04:23:32 +0000 https://www.fedai.org/?p=1733


From November 7th of 2019, Artificial intelligence and artistic creation exploration exhibition was held in Shanghai Ming Contemporary Art Museum (McaM). A lot of experts and the artistic pioneers came here to present an artificial intelligence art visual feast for the audience by the artistic works, exploring the infinite possibilities of “AI + Artistic”.

Unlike the usual exhibits, this art exhibition also has one special exhibit. The audiences not only can see it, but they also can play it. That is the visualization of AI games based on federated learning (Federated Learning Demo). Federated Learning Demo can enable the audiences to participate in the process of AI modeling, using the robot trained by ourselves to fight with the game robot. With interactivity and enjoyment, teaching in fun make the understanding of AI to be easy and interesting.

The federated learning demo is a semi-interactive racing game. The audiences can participant in the game so that they can understand the technical theory and advantage of federated learning AI visually. As we all know, in the racing game, the traditional gameplay is trying, again and again, to get more familiar with the map route in order to achieve the maximum speed. Due to the personal energy and the limited opportunities (perhaps limited to the talent), the players can’t make ensure to win after several times of training. Playing on solo, the traditional way is a kind of the AI model relying on the training by a single player. This AI model’s performance is general because the model just learns a little data from one player. However, in this process, federated learning technology can bypass the players’ data and directly combine the AI models trained by the data of every player, train and improve the AI model of federated learning constantly, and become the final winner. In other words, no matter how the single-player good is, he/she can only fight alone, while federated learning can win with multiple parties.

Above: Training process of AI

Below: Player interface

It was reported that the Shanghai art exhibition isn’t the first appearance for federated learning demo. As the first gamification semi-interactive demo of federated learning in China, in August it was also exhibited at the International Joint Conferences on Artificial Intelligence (IJCAI 2019) held in Macao. From the academic conference to the art exhibition, federated learning demo not only can be participated and visible but also it is gamification, which can enable audiences to join into the operation of federated learning technology. Providing a new path for AI to appear in public, federated learning demo also lower the threshold of AI perception for non-technical people.

The plenty of data used for model training during the game also reminds people of the data privacy and security problems behind the game. If we don’t share the data, that would be data silos. It is hard to train an effective AI model like practicing alone. Who can protect user privacy and security if we share data? Federated learning technology is a good demonstration – sharing the data, not the underlying data, which can enable data to never leave local storage and jointly modeling so as to improve the performance of machine learning. Thus, it also protects the data privacy and security to provide a new idea about solving the contradiction of data regulation, data soils and the development of AI.

As a pioneer and promoter in the development of Federated Learning technology in China, WeBank keeps promoting the application implementation of federated learning in different industries. For example, in the computer vision industry, WeBank and Extreme Vision jointly launch the “federated visual system”, helping the company expand the data application scope to share the success of data models. Meanwhile, you don’t need to concern the problem of data security. Except for the computer vision field, federated learning technology has also applied in finance, medicine, city management and other fields to help the intelligence upgrade of industries.

It can be predicted that in the future, the protection of data security and data privacy will be more concerned. Federated learning will apply in more industries and scenarios, accelerating the arrival of the intelligent future. In the path to the intelligent future, we not only need hard technology like federated learning, but we should also own an interesting tool, like the federated learning demo to help us understand the new technology easily. Only in this way can everyone participate and benefit from the wisdom of the future. As the demo creator Quan Li and Xiguang Wei said, “Whether it’s technology or the tools to explain it, it has to focus on human so that people can live a better life at a lower cost.”