Centraltotheconstruction of our benchmark is a novel semi-automatic crowdsourced pipeline which uses optical flow estimates to compensate for easier, short-term motion like camera shake,allowing annotators tofocus onharder sections ofvideo.

Forinstance,comparedto existing solutions, we showthat SCLTS plays the (non-optimal) baseline action at most O(logT) times (compared toO( T)). Finally, we make connections to another studied form of "safety constraints" that takes the form of anupper bound on the instantaneous reward.

However, theresultsofKhezeliand Bitar(2019) have notbeenextendedtothiscase; 4) In Section3.1,

A promising direction istouse linear recurrent architectures (LRUs), where dense recurrent weights are replaced with a complex-valued diagonal, making RTRL efficient.