Inspired by a first-order optimization method called mirror descent, this paper proposes topredict asequence ofrewardfunctions, which areiterativesolutions for a constrained convex problem. IRL solutions derived by mirror descent are tolerant totheuncertainty incurred bytargetdensity estimation sincetheamount of reward learning is regulated with respect to local geometric constraints.

For eachDk subsets, respectively, we follow the same subsampling procedure used in the singlevariate case. That is, we letq10 = 10% of the first data subset andq20 = 10% of the second data subset.

Recently,Sutton[23]proposed anewview on action selection.

Traditional robust reinforcement learning often depends on rectangularity assumptions, where adverse probability measures of outcome states are assumed to be independent across different states and actions.

While many models have been developed for characterizing behavior in binary decision-making and bandit tasks, comparatively little work has focused onanimal decision-making inmorecomplextasks,suchasnavigationthrough a maze.

Many potential applications of reinforcement learning (RL) have intractably-large state spaces.

In the specific case of MuZero Reanalyse it also performs a tree search rather than a single step lookaheadusedinDyna.