Shanghai, China – Ant Group announced three initiatives to make trusted AI more accessible. This includes making its graph processing database TuGraph open source; launching an industrial-grade AI security testing platform; and unveiling a privacy-preserving computation open platform. Announced at the World Artificial Intelligence Conference (“WAIC”) in Shanghai, these actions aim to make trusted AI solutions more accessible to enable industrial multi-party collaborations in the digital economy.
A growing number of industries have been adopting AI technology at a fast pace, which has significantly increased demand for trusted multi-party collaboration solutions and AI systems that are secure, reliable and robust.
“We believe that privacy-preserving computation, blockchain, graph-processing technology, distributed database and green computing are the fundamental technologies for a large-scale industrial application of AI in the digital economy. These fundamental technologies are critical for overcoming challenges arising from the surging demand for security and trust in multi-party industrial collaboration, surging need for data storage and energy consumption, as well as the rising complexity in managing relationships among different parties,” says Tao Wei, vice president and chief technology security officer at Ant Group. “Ant Group has been continuously investing in the research and development of these technologies to facilitate more industrial collaborations in the digital economy.”
As part of the actions, Ant Group has also made TuGraph open source for global developers on Github. TuGraph is one of the key components in the company’s large-scale graph-processing system and was jointly developed by Ant Group and Tsinghua University. The graph-processing technology has a wide range of applications, such as risk control, anti-fraud and knowledge graph.
TuGraph was awarded the World Leading Internet Scientific and Technology Achievements in 2021, and broke the throughput record in the audited results of the LDBC Social Network Benchmark (Interactive Workload) in August 2022.
Ant Group, China Academy of Information and Communications Technology (“CAICT”) and Tsinghua University also launched an industrial-grade AI security testing platform to provide developers a one-stop evaluation solution from adversarial tests to defense and reinforcements. This platform also helps developers recognise and predict model risks through a single click, building safer and more robust AI models and detect systemic risks in advance. The platform is embedded with independently developed adversarial-AI security computing engines, which can improve the computing performance by dozens of times. Meanwhile, the platform now offers robustness evaluation for global developers with more functions planned in the pipeline.
In July 2022, Ant Group made its privacy-preserving computation framework open source. The privacy computing platform released this time features a one-stop solution for industrial users to create privacy computation applications in various use cases, such as medical service, insurance claims, joint marketing and risk control. It also comes with the additional option to build applications without writing a single line of code.
Follow us and Comment on Twitter @TheEE_io