Two-time-scale optimization is a framework introduced in Zeng et al. (2024) that abstracts a range of policy evaluation and policy optimization problems in reinforcement learning (RL). Akin to bi-level optimization under a particular type of stochastic oracle, the two-time-scale optimization framework has an upper level objective whose gradient evaluation depends on the solution of a lower level problem, which is to find the root of a strongly monotone operator. In this work, we propose a new method for solving two-time-scale optimization that achieves significantly faster convergence than the prior arts. The key idea of our approach is to leverage an averaging step to improve the estimates of the operators in both lower and upper levels before using them to update the decision variables. These additional averaging steps eliminate the direct coupling between the main variables, enabling the accelerated performance of our algorithm. We characterize the finite-time convergence rates of the proposed algorithm under various conditions of the underlying objective function, including strong convexity, convexity, Polyak-Lojasiewicz condition, and general non-convexity. These rates significantly improve over the best-known complexity of the standard two-time-scale stochastic approximation algorithm. When applied to RL, we show how the proposed algorithm specializes to novel online sample-based methods that surpass or match the performance of the existing state of the art. Finally, we support our theoretical results with numerical simulations in RL.

An October 2023 crash between a GM Cruise robotaxi and a pedestrian in San Francisco resulted not only in a severe injury, but also dramatic upheaval at that company that will likely have lasting effects throughout the industry. The issues stem not just from the crash facts themselves, but also how Cruise mishandled dealing with their robotaxi dragging a pedestrian under the vehicle after the initial post-crash stop. A pair of external investigation reports provide raw material describing the incident and critique the company's response from a regulatory interaction point of view, but did not include potential safety recommendations in scope. We use that report material to highlight specific facts and relationships between events by tying together different pieces of the report material. We then explore safety lessons that might be learned with regard to technology, operational safety practices, and organizational reaction to incidents.

Mumbai-based startup Qure uses an AI imaging tool qER to save precious minutes for emergency room staff to take action based on head CT scans. After deployment in India and several other countries, qER is now entering the US where 75 million CT scans are performed every year. A couple of weeks ago, Qure received US FDA 510 (k) clearance for this product. What makes it special is a four-in-one clearance. The tool has been cleared for triaging four critical conditions--intracranial bleeds, mass effect (due to spaces in the brain filling up), midline shift (in the brain's alignment), and cranial fractures.