Many a cinephile has asked themselves the question: What if Andrei Tarkovsky's Solaris, the avant-garde 1972 sci-fi classic about a widowed space explorer forced to grapple with his grief while on a mission to a mysterious planet, starred Adam Sandler and Carey Mulligan as a fracturing married couple, alongside Paul Dano as the voice of a giant benevolent space spider? And what if Isabella Rossellini were the leader of Czechoslovakia's space program, which somehow, in this universe's alternate version of political and technological history, was the best-equipped in the world to send a manned mission to the outer reaches of Jupiter? The result of that mashup might be something like Spaceman, an oddball psychological drama from the Swedish director Johan Renck, best known for a long résumé of music videos and lately for helming all five episodes of the acclaimed HBO miniseries Chernobyl. The script, adapted by Colby Day from the 2017 novel Spaceman of Bohemia by Jaroslav Kalfar, leaves many questions unanswered. If the mission of Sandler's character--the depressive, remote, and career-obsessed Jakub--is so significant to humanity's future, it isn't clear why he would have been sent into space all alone.

We propose the first boosting algorithm for off-policy learning from logged bandit feedback. Unlike existing boosting methods for supervised learning, our algorithm directly optimizes an estimate of the policy's expected reward. We analyze this algorithm and prove that the excess empirical risk decreases (possibly exponentially fast) with each round of boosting, provided a ''weak'' learning condition is satisfied by the base learner. We further show how to reduce the base learner to supervised learning, which opens up a broad range of readily available base learners with practical benefits, such as decision trees. Experiments indicate that our algorithm inherits many desirable properties of tree-based boosting algorithms (e.g., robustness to feature scaling and hyperparameter tuning), and that it can outperform off-policy learning with deep neural networks as well as methods that simply regress on the observed rewards.

Adam Sandler is an absolute talent in the entertainment business and a comic genius. Sandler has a lengthy, excellent résumé that spans more than three decades, and he has consistently demonstrated his dominance in his field. Have you seen his specials from his stand-up comedy? Audiences love Sandler's distinct delivery and flair, which never fails to make them laugh. His talent, though, extends beyond.

Practical adoption of artificial intelligence (AI) faces a variety of roadblocks--splashy, high-profile deployments of AI have not been received well, with Microsoft's "Tay" bot on Twitter parroting anti-Semetic vitriol just 16 hours after launch. Similarly, Amazon's AI-powered hiring process displayed bias against women and the company marketed unreliable facial recognition technology to municipal law enforcement. AI often reflects the biases--including, and especially, unconscious biases--of the designers, which would make Facebook attempting to build an AI with an "ethical compass" a concerning prospect, given the multitude of other problems the social network has experienced. This is a problem that necessarily requires diversity of thought, according to Northeastern University's Ethics Institute and professional services firm Accenture, which published a guide to building data and AI ethics committees. Such committees are, by definition, not achievable by pooling together people of similar backgrounds to debate the merits of AI design.

Amazon Inc. (AMZN) is on a roll. The world's biggest online retailer blew away analyst expecations Thursday with a 513 million profit, its fourth straight profitable quarter, sending its stock up nearly 12 percent after hours on the Nasdaq. The profit number for the first quarter of 2016 swung from a net loss of 57 million in the same quarter last year and nearly doubled an estimate of 272.6 million, according to analysts polled by Thomson Reuters. The e-commerce giant posted 29.13 billion in revenue, up by 28 percent from 22.7 billion a year ago, driven by growth in its product sales and cloud computing businesses. Those figures beat analyst expectations of 27.9 billion in revenue.