 |
| A.I. Phoenix taps federated learning for novel collaborative AI applications |
|
 |
Federated learning is a new kid on the AI machine learning block. A.I. Phoenix, a Cyberport Creative Micro Fund (CCMF) grantee, is leveraging this powerful technology to connect – without exchanging – multiple decentralised data samples. Its customised data mining, processing, and analysis solutions can vastly expand the range and power of data available to customers, so as to deliver exceptional efficiency and value without requiring much investments in acquiring new data sets or infrastructures.
For example, banks that want to reactivate dormant accounts can avail themselves of A.I. Phoenix’s services. Instead of having investment advisors calling customers one by one, A.I. Phoenix’s system can help the advisors understand their customers better by developing more accurate customer portraits. Using the “transfer learning” technique, the system can combine the bank’s financial data with the non-financial data of external parties and rank the customers by their probability and interest in buying new products or services. This way, the advisors can focus on the top-ranked customers and improve efficiency and conversion rate.
All of this does not require data to change hands because, instead of combining the internal and external data sets before doing the analyses, transfer learning breaks machine learning analyses into sub-tasks, trains the models with each data set locally, and then collects and combines all of the trained models in the central server for analysis. During the whole process, data stay in each party’s local servers, and all data sets are used without compromising data privacy.
The beauty of A.I. Phoenix’s solutions is that they can be applied to multiple industries, and the start-up is already serving clients with different business requirements. For example, A.I. Phoenix is providing intelligent user management tools to retail banking departments, intelligent online course matching and marketing empowerment for online one-to-one English tutorials, and AI analysis solutions for Mainland government HR departments.
Intent on starting with a bang, A.I. Phoenix joined a host of start-up competitions across China soon after starting the company and won dozens of accolades. It attracted the attention of nearly 70 investors, government agencies, and incubators in more than 30 cities, and has since been in talks with such investors as GoldenSand Capital, Shenzhen Venture Capital, and Zhuhai Hi-Tech Venture Capital.
A.I. Phoenix has already set up presence in Qianhai (Shenzhen) last year and is looking to land in more major Mainland cities this year including Beijing, Shanghai, Guangzhou, and Hangzhou. The start-up is seeking to capitalise on the strengths and advantages of different cities to take its business forward and to foster win-win collaborations for Hong Kong, Macau, and Mainland China.
. |
|
|
|
|
|