“…It usually uses learning methods to extract information representing user characteristics from different types of sensors on mobile devices to build a model. Such as face [6,7], fingerprint [8,9], voice [10,11], environmental location [12,13], keystroke behavior [14,15], finger movement [16,17], etc. However, the above-mentioned biometric information collection usually requires invoking the privacy-related permissions of the mobile device, which makes users worry that their privacy-related information may be leaked, and cannot take into account the security, privacy, and usability requirements of mobile authentication jointly at the same time.…”