An Approach to Dynamically Integrate Heterogeneous AI Components in Multimodal User Authentication - tinyML Asia 2021

Explore an innovative approach to dynamically integrate heterogeneous AI components in a multimodal user authentication system during this 23-minute conference talk from tinyML Asia 2021. Delve into the challenges of combining multiple AI-backed authentication methods, such as facial recognition, voice recognition, and touch patterns, each employing different AI technologies. Learn about the development of a common interface and domain-specific language (DSL) for component integration, and discover how a framework ensures minimal performance requirements are met while optimizing component invocation strategies. Gain insights into the use of simulators for evaluating integration program performance and the potential for combining human and artificial intelligence in strategy development. Understand the application of this approach in building the next-generation user authentication system, DZ Security, and its implications for similar systems in the field of tinyML.

Intro
Application Domain Continuous User Authentication
Basic System Structure
Decompose the System
Build the Foundation
Focus on the Core
Putting it All Together