While this setup has enjoyed a lot of attention in the applied community, there hasn't be theoretical work that even formalizes the desired guarantees.

However, because they are not identifiable without any assumptions, various assumptions have been utilized to evaluate the joint probabilities of potential outcomes, e.g., the assumption of monotonicity (Pearl, 2009; Tian and Pearl, 2000), the independence between potential outcomes (Robins and Richardson, 2011), the condition of gain equality (Li and Pearl, 2019), and the specific functional relationshipsbetween cause and effect (Pearl, 2009). Unlike existing identification conditions, in order to evaluate the joint probabilities of potential outcomeswithoutsuch assumptions,this paper proposestwo types of novel identification conditions using covariate information. In addition, when the joint probabilities of potential outcomes are identifiable through the proposed conditions, the estimation problem of the joint probabilities of potential outcomes reduces to that of singular models and thus they can not be evaluated by standard statistical estimation methods. To solve the problem,this paper proposes a new statisticalestimationmethod based on the augmented Lagrangianmethod and shows the asymptoticnormality of the proposed estimators. Given space constraints, the proofs, the details on the statistical estimationmethod, some numerical experiments, and the case study are provided in the supplementary material.

Contextual multinomial logit (MNL) bandits capture many real-world assortment recommendation problems such as online retailing/advertising. However, prior work has only considered (generalized) linear value functions, which greatly limits its applicability. Motivated by this fact, in this work, we consider contextual MNL bandits with a general value function class that contains the ground truth, borrowing ideas from a recent trend of studies on contextual bandits. Specifically, we consider both the stochastic and the adversarial settings, and propose a suite of algorithms, each with different computation-regret trade-off. When applied to the linear case, our results not only are the first ones with no dependence on a certain problem-dependent constant that can be exponentially large, but also enjoy other advantages such as computational efficiency, dimension-free regret bounds, or the ability to handle completely adversarial contexts and rewards.

'Picasso,' we explore the potential for unique, crowd-sourced styles that could provide deep insights into the collective creative psyche of users

Is surprise box-office hit Iron Lung the future of'video game films'? The YouTube gaming star's weird and divisive adaptation of his obscure horror film is a game within a film about a game - and hints at new directions for storytelling Don't get Pushing Buttons delivered to your inbox? S omething weird struck me early on while watching the movie Iron Lung, which has so far taken $32m at the box office, despite being a grungy low-budget sci-fi thriller adapted from an independent video game few people outside of the horror gaming community have even heard of. Set after a galactic apocalypse, it follows a convict who must buy his freedom by piloting a rusty submarine through an ocean of human blood on a distant planet. Ostensibly, he's looking for relics that may prove vital for scientific research, but what he finds is much more ghastly.