As we increasingly rely on biometric data such as fingerprints and facial recognition to secure our digital lives, many of us are reaching a point where using strings of characters such as passwords to encrypt our data will become a thing of the past - and perhaps quite soon. Big tech seems to agree. The triumvirate of Apple, Microsoft and Google all recently signed up to FIDO2 WebAuthn, a standard created by the Fast IDentity Online Alliance (FIDO) - an industry association which aims to shift our reliance away from passwords in favor of text-free alternatives. And according to Eve Maler, CTO at access management firm ForgeRock, passwords are no longer a viable option for the security conscious, with a number of advanced technologies looking to replace them. The headline act is behavioral biometrics, which has, "a significant role to play in making passwordless security a reality", according to Maler.

We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 - 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Identity and access management (IAM) provider ForgeRock recently held its annual IDLive conference in Austin, Texas. One of the most compelling sessions involved ForgeRock CTO Eve Maler, who discussed the future of IAM and how it's now being heavily infused with artificial intelligence (AI) to make it more effective. The future that Maler described is very much aligned with the company's mission to "help people safely and simply access the connected world" and its vision of "never having to log in again." While IAM has historically been a part of the IT plumbing to manage employee access within companies, it has emerged as a technology with a significant impact on all users -- employees, consumers, citizens and others -- in the new post-pandemic digital world that is evolving into Web3.

In this paper, we study adaptive online convex optimization, and aim to design a universal algorithm that achieves optimal regret bounds for multiple common types of loss functions. Existing universal methods are limited in the sense that they are optimal for only a subclass of loss functions. To address this limitation, we propose a novel online method, namely Maler, which enjoys the optimal $O(\sqrt{T})$, $O(d\log T)$ and $O(\log T)$ regret bounds for general convex, exponentially concave, and strongly convex functions respectively. The essential idea is to run multiple types of learning algorithms with different learning rates in parallel, and utilize a meta algorithm to track the best one on the fly. Empirical results demonstrate the effectiveness of our method.