We present tools for the analysis of Follow-The-Regularized-Leader (FTRL), Dual Averaging, and Mirror Descent algorithms when the regularizer (equivalently, proxfunction or learning rate schedule) is chosen adaptively based on the data. Adaptivity can be used to prove regret bounds that hold on every round, and also allows for data-dependent regret bounds as in AdaGrad-style algorithms (e.g., O...

The current study sought to answer three key questions about explaining the emotion of regret in the domain of casual sex: Are sex differences in sexual regret robust or attenuated in a highly egalitarian culture? What proximate psychological variables might explain sex differences in sexual regret? And what accounts for within-sex variation in experiences of sexual regret about casual sex. We ...

What do people think about the emotion of regret? Recent demonstrations of the psychological benefits of regret have been framed against an assumption that most people find regret to be aversive, both when experienced but also when recalled later. Two studies explored lay evaluations of regret experiences, revealing them to be largely favorable rather than unfavorable. Study 1 demonstrated that...

We present a reduction from cost sensitive classi cation to binary classi cation based on (a modi cation of) error correcting output codes. The reduction satis es the property that regret for binary classi cation implies l2-regret of at most 2 for cost-estimation. This has several implications: 1) Any regret-minimizing online algorithm for 0/1 loss is (via the reduction) a regret-minimizing onl...

"Regret aversion" is proposed to explain a tendency to avoid future choices that have induced past regret. However, regret might also motivate us to repeat previous regret-related choices to make up for their previous selection, a behavior resembling "chasing" in the context of gambling. In the current experiment, we acquired fMRI brain data while participants placed monetary bets on repeated g...

We present tools for the analysis of Follow-The-Regularized-Leader (FTRL), Dual Averaging, and Mirror Descent algorithms when the regularizer (equivalently, proxfunction or learning rate schedule) is chosen adaptively based on the data. Adaptivity can be used to prove regret bounds that hold on every round, and also allows for data-dependent regret bounds as in AdaGrad-style algorithms (e.g., O...

This commentary is about (Chiang et al., 2012b). This paper is the result of a merge between two papers, (Yang et al., 2012) and (Chiang et al., 2012a). Both papers address the same question: is it possible to obtain regret bounds in various online learning settings that depend on some notion of variation in the costs, rather than the number of periods? Both papers give remarkably similar algor...

We consider the framework of stochastic multi-armed bandit problems and study the possibilities and limitations of forecasters that perform an on-line exploration of the arms. A forecaster is assessed in terms of its simple regret, a regret notion that captures the fact that exploration is only constrained by the number of available rounds (not necessarily known in advance), in contrast to the ...

We present a new anytime algorithm that achieves near-optimal regret for any instance of finite stochastic partial monitoring. In particular, the new algorithm achieves the minimax regret, within logarithmic factors, for both “easy” and “hard” problems. For easy problems, it additionally achieves logarithmic individual regret. Most importantly, the algorithm is adaptive in the sense that if the...

We search for behavioral rules that attain minimax regret under geometric discounting in the context of repeated decision making in a stationary environment where payo¤s belong to a given bounded interval. Rules that attain minimax regret exist and are optimal for Bayesian decision making under the prior where learning can be argued to be most di¢cult. Minimax regret can be attained by randomiz...