Y. J. Ma | Searcharxiv

arXiv2022

Regret Bounds for Risk-Sensitive Reinforcement Learning

In safety-critical applications of reinforcement learning such as healthcare and robotics, it is often desirable to optimize risk-sensitive objectives that account for tail outcomes rather than expected reward. We prove the first regret bounds for reinforcement learning under a general class of risk-sensitive objectives including the popular CVaR objective. Our theory is based on a novel characterization of the CVaR objective as well as a novel optimistic MDP construction.

cs.LG↗

arXiv2007

Candidate chiral doublet bands in the odd-odd nucleus $^{126}$Cs

The candidate chiral doublet bands recently observed in $^{126}$Cs have been extended to higher spins, several new linking transitions between the two partner members of the chiral doublet bands are observed, and $γ$$-$intensities related to the chiral doublet bands are presented by analyzing the $γ$$-$$γ$ coincidence data collected earlier at the NORDBALL through the $^{116}$Cd$($$^{14}$N, 4n$)$$^{126}$Cs reaction at a beam energy of 65 MeV. The intraband $B(M1)/B(E2)$ and interband $B(M1)_{in}/B(M1)_{out}$ ratios and the energy staggering parameter, S(I), have been deduced for these doublet bands. The results are found to be consistent with the chiral interpretation for the two structures. Furthermore, the observation of chiral doublet bands in $^{126}$Cs together with those in $^{124}$Cs, $^{128}$Cs, $^{130}$Cs and $^{132}$Cs also indicates that the chiral conditions do not change rapidly with decreasing neutron number in these odd-odd Cesium isotopes.

nucl-ex↗