Continuous authentication
Identity in every step.
The challenge
Passwords, PINs and smart cards verify a person once. After that, anyone holding the device is trusted.
Our approach
We verify identity continuously and silently, from the way a person walks and from their body's signals. Our neural networks stop computing as soon as they're confident, so they run on low-power processors in watches, phones and field equipment.
Highlights
- Gait-based identity from motion sensors already built into everyday devices
- Runs on low-power processors with no GPU
- Fusion of gait with ECG, skin response (GSR), and muscle and brain signals
- Designed to move from software to FPGA and chip-level implementations
Built on
Early-exit neural networks