An important property for lifelong-learning agents is the ability to combine existing skills to solve unseen tasks. In general, however, it is unclear how to compose skills in a principled way. We provide a "recipe" for optimal value function composition in entropy-regularised reinforcement learning (RL) and then extend this to the standard RL setting. Composition is demonstrated in a video game environment, where an agent with an existing library of policies is able to solve new tasks without the need for further learning.

Buildings are a central part of the world today. We live, work and play in them. We socialize, learn and engage in them. In fact, it's estimated that we spend 93% of our time in buildings. Of course, buildings don't remain static once the builder hands over the keys – their uses, occupants and components vary and change with time.

There is a new and apparently better way to fix grainy photos, and it uses AI. Artificial Intelligence, one must say, coupled with lots of computing power, but that's growing day after day. Artificial Intelligence and machine learning will never cease to surprise us, apparently. Each new day, there's some new announcement, covering different fields: AI can render 3D hair in real time, smell illnesses in human breath, assess infrastructure quality in Africa or help transform audio into music playing avatars. Now it can also help photographers get rid of noise in their photos.

Herein, we present a system for hyperspectral image segmentation that utilizes multiple class--based denoising autoencoders which are efficiently trained. Moreover, we present a novel hyperspectral data augmentation method for labelled HSI data using linear mixtures of pixels from each class, which helps the system with edge pixels which are almost always mixed pixels. Finally, we utilize a deep neural network and morphological hole-filling to provide robust image classification. Results run on the Salinas dataset verify the high performance of the proposed algorithm.

Due to the iterative nature of most nonnegative matrix factorization (\textsc{NMF}) algorithms, initialization is a key aspect as it significantly influences both the convergence and the final solution obtained. Many initialization schemes have been proposed for NMF, among which one of the most popular class of methods are based on the singular value decomposition (SVD). However, these SVD-based initializations do not satisfy a rather natural condition, namely that the error should decrease as the rank of factorization increases. In this paper, we propose a novel SVD-based \textsc{NMF} initialization to specifically address this shortcoming by taking into account the SVD factors that were discarded to obtain a nonnegative initialization. This method, referred to as nonnegative SVD with low-rank correction (NNSVD-LRC), allows us to significantly reduce the initial error at a negligible additional computational cost using the low-rank structure of the discarded SVD factors. NNSVD-LRC has two other advantages compared to previous SVD-based initializations: (1) it provably generates sparse initial factors, and (2) it is faster as it only requires to compute a truncated SVD of rank $\lceil r/2 + 1 \rceil$ where $r$ is the factorization rank of the sought NMF decomposition (as opposed to a rank-$r$ truncated SVD for other methods). We show on several standard dense and sparse data sets that our new method competes favorably with state-of-the-art SVD-based initializations for NMF.

In this profile series, we interview AI innovators on the front-lines - those who have dedicated their life's work to improving the human condition through technology advancements. Born and raised in Ethiopia, Gebru immigrated to the US at 16 to earn her PhD from Stanford Artificial Intelligence Laboratory and just finished her year as a post-doctoral researcher at Microsoft Research in New York. While she was still a PhD student, she co-founded Black in AI, an organization fostering collaboration and discussing initiatives to increase the representation of Black people in the field. Was there a moment where you questioned your path? I mean, when I first did analog circuit design, I was very much into hardware and that was my main focus while at Apple.

This paper aims to instigate discussion, answering a call for papers from the ICML Machine Learning Debates workshop. While we stand by the points represented here, we do not purport to offer a full or balanced viewpoint or to discuss the overall quality of science in ML. In many aspects, such as reproducibility, the community has advanced standards far beyond what sufficed a decade ago. We note that these arguments are made by us, against us, by insiders offering a critical introspective look, not as sniping outsiders. The ills that we identify are not specific to any individual or institution. We ourselves have fallen into these patterns, and likely will again in the future. Exhibiting one of these patterns doesn't make a paper bad nor does it indict the paper's authors, however we believe that all papers could be made stronger by avoiding these patterns. While we provide concrete examples, our guiding principles are to (i) implicate ourselves, and (ii) to preferentially select from the work of better-established researchers and institutions that we admire, to avoid singling out junior students for whom inclusion in this discussion might have consequences and who lack the opportunity to reply symmetrically. We are grateful to belong to a community that provides sufficient intellectual freedom to allow us to express critical perspectives. In each subsection below, we (i) describe a trend; (ii) provide several examples (as well as positive examples that resist the trend); and (iii) explain the consequences. Pointing to weaknesses in individual papers can be a sensitive topic. To minimize this, we keep examples short and specific.

How is AI changing data science and some of the potential applications? In my last post, I talked to Prenton Chetty, Head of Analytics at Nedbank, about the impact of analytics on the business, and how modelling has changed the conversation and improved decision making. This post continues our discussion, and explores Prenton's views on how data science is changing with AI, and some of the potential applications. More and more, the business was asking us to solve problems that just could not be touched using traditional methods. There were situations where we didn't have the right tools, so we needed something new.

Moving stuff by air may be quick and convenient, but it's also horridly expensive, accounting for just 1 percent of global shipping by volume--and 35 percent of it by cost. So while autonomous drones dropping a few pounds of snacks or medical supplies are generating plenty of buzz, two Bulgarian brothers see an opening in the long-haul business. And they think they've got the tech to start flying hundreds of pounds of cargo over hundreds of miles, no pilot or 747 required. Svilen and Konstantin Rangelov are the CEO and chief technology officer, respectively, of Dronamics. They've spent the past four years developing an aircraft that can haul nearly 800 pounds of cargo up to 1,550 miles, a far cry from the 10 or 15 miles, or even just a few blocks, that most drone delivery services are targeting.

Better Life Lab is a partnership of Slate and New America. In an age where every day brings more doomsday forecasts of massive technologicallybdriven unemployment, from driverless cars to A.I. robots as caregivers, journalist Annie Lowrey set out to answer a question: Is it possible to live in a world where we get what she calls "wages for breathing"? This week her findings come out in Give People Money: How a Universal Basic Income Would End Poverty, Revolutionize Work, and Remake the World. We spoke about what the idea of giving every American cash--no strings attached--would mean for work, gender inequality, and American identity, and whether it's actually a policy that could pass in the U.S. given the current climate of tying even the most basic benefits to paid work. This interview has been condensed and edited for clarity.