Drones have a fundamental design problem. The kind of drone that can carry large payloads at high speeds over long distances is fundamentally different from the kind of drone that can take off and land from a small area. In very simple terms, for the former, you want fixed wings, and for the latter, you want rotors. This problem has resulted in a bunch of weird drones that try to do both of these things at once, usually by combining desired features from fixed-wing drones and rotorcraft. We've seen tail-sitter drones that can transition from vertical take off to horizontal flight; we've seen drones with propeller systems that swivel; and we've seen a variety of airframes that are essentially quadrotors stapled to fixed-wing aircraft to give them vertical take-off and landing capability.

PwC has highlighted six important factors for healthcare organisations to effectively incorporate AI in a new report. The report, 'From Virtual to Reality: Six imperatives for becoming an AI-ready healthcare business', outlines use cases and learnings from six main areas, including Leadership & Culture, Workforce Transformation, Clinical Effectiveness, Commercial Investment, Public Readiness and Regulation, Ethics and Confidentiality. PwC outlines the importance of several factors for enabling AI-powered healthcare businesses, including creating the right leadership and culture; improving clinical effectiveness; investing in and funding businesses and technology; ensuring appropriate regulation and addressing concerns about ethics and confidentiality. The company also says that the Middle East is in a unique position to lead the development of international standards and become a hub for AI research and development. Factors such as the proactive development of governance and regulatory frameworks will facilitate the implementation of AI in healthcare.

Without doubt, artificial intelligence and machine learning are major areas of innovation for the greater tech community. If you are a tech freelancer and eager to stay in touch with future directions, you will want to know what companies like Google are investing in, the new technologies they are advancing and the research priorities they are supporting or sponsoring. And, you are in luck! A post yesterday by Jeff Dean, senior fellow and Google AI lead, on behalf of the Google Research Community, reviews how Google has focused its research talent and dollars. I've provided a thumbnail summary of the priorities, quoting descriptions from the blog post.

Modern human DNA computational analysis suggests that the extinct species was a hybrid of Neanderthals and Denisovans and cross bred with Out of Africa modern humans in Asia. This finding would explain that the hybrid found this summer in the caves of Denisova -- the offspring of a Neanderthal mother and a Denisovan father -- was not an isolated case, but rather was part of a more general introgression process. The study, published in Nature Communications, uses deep learning for the first time ever to account for human evolution, paving the way for the application of this technology in other questions in biology, genomics and evolution. One of the ways of distinguishing between two species is that while both of them may cross breed, they do not generally produce fertile descendants. However, this concept is much more complex when extinct species are involved.

The gradient noise (GN) in the stochastic gradient descent (SGD) algorithm is often considered to be Gaussian in the large data regime by assuming that the classical central limit theorem (CLT) kicks in. This assumption is often made for mathematical convenience, since it enables SGD to be analyzed as a stochastic differential equation (SDE) driven by a Brownian motion. We argue that the Gaussianity assumption might fail to hold in deep learning settings and hence render the Brownian motion-based analyses inappropriate. Inspired by non-Gaussian natural phenomena, we consider the GN in a more general context and invoke the generalized CLT (GCLT), which suggests that the GN converges to a heavy-tailed $\alpha$-stable random variable. Accordingly, we propose to analyze SGD as an SDE driven by a L\'{e}vy motion. Such SDEs can incur `jumps', which force the SDE transition from narrow minima to wider minima, as proven by existing metastability theory. To validate the $\alpha$-stable assumption, we conduct extensive experiments on common deep learning architectures and show that in all settings, the GN is highly non-Gaussian and admits heavy-tails. We further investigate the tail behavior in varying network architectures and sizes, loss functions, and datasets. Our results open up a different perspective and shed more light on the belief that SGD prefers wide minima.

A novel Gromov-Wasserstein learning framework is proposed to jointly match (align) graphs and learn embedding vectors for the associated graph nodes. Using Gromov-Wasserstein discrepancy, we measure the dissimilarity between two graphs and find their correspondence, according to the learned optimal transport. The node embeddings associated with the two graphs are learned under the guidance of the optimal transport, the distance of which not only reflects the topological structure of each graph but also yields the correspondence across the graphs. These two learning steps are mutually-beneficial, and are unified here by minimizing the Gromov-Wasserstein discrepancy with structural regularizers. This framework leads to an optimization problem that is solved by a proximal point method. We apply the proposed method to matching problems in real-world networks, and demonstrate its superior performance compared to alternative approaches.

Abstract: Machine learning competitions such as those organized by Kaggle or KDD represent a useful benchmark for data science research. In this work, we present our winning solution to the Game Data Mining competition hosted at the 2017 IEEE Conference on Computational Intelligence and Games (CIG 2017). The contest consisted of two tracks, and participants (more than 250, belonging to both industry and academia) were to predict which players would stop playing the game, as well as their remaining lifetime. The data were provided by a major worldwide video game company, NCSoft, and came from their successful massively multiplayer online game Blade and Soul. Here, we describe the long short-term memory approach and conditional inference survival ensemble model that made us win both tracks of the contest, as well as the validation procedure that we followed in order to prevent overfitting. In particular, choosing a survival method able to deal with censored data was crucial to accurately predict the moment in which each player would leave the game, as censoring is inherent in churn. The selected models proved to be robust against evolving conditions--since there was a change in the business model of the game (from subscription-based to free-to-play) between the two sample datasets provided--and efficient in terms of time cost. Thanks to these features and also to their ability to scale to large datasets, our models could be readily implemented in real business settings. Keywords: churn; competition; video games; user behavior; behavioral data 1.

Domain adaptation has become a prominent problem setting in machine learning and related fields. This review asks the questions: when and how a classifier can learn from a source domain and generalize to a target domain. As for when, we review conditions that allow for cross-domain generalization error bounds. As for how, we present a categorization of approaches, divided into, what we refer to as, sample-based, feature-based and inference-based methods. Sample-based methods focus on weighting individual observations during training based on their importance to the target domain. Feature-based methods focus on mapping, projecting and representing features such that a source classifier performs well on the target domain and inference-based methods focus on alternative estimators, such as robust, minimax or Bayesian. Our categorization highlights recurring ideas and raises a number of questions important to further research.

Deep Reinforcement Learning (DRL) has become increasingly powerful in recent years, with notable achievements such as Deepmind's AlphaGo. It has been successfully deployed in commercial vehicles like Mobileye's path planning system. However, a vast majority of work on DRL is focused on toy examples in controlled synthetic car simulator environments such as TORCS and CARLA. In general, DRL is still at its infancy in terms of usability in real-world applications. Our goal in this paper is to encourage real-world deployment of DRL in various autonomous driving (AD) applications. We first provide an overview of the tasks in autonomous driving systems, reinforcement learning algorithms and applications of DRL to AD systems. We then discuss the challenges which must be addressed to enable further progress towards real-world deployment.

She begins at Zoox on Feb. 26. Her hiring makes Ms. Evans, born in Senegal and raised in Paris, one of the most high-profile black women running a Silicon Valley tech company. She succeeds Tim Kentley-Klay, the co-founder who was removed as CEO by Zoox's board last August after the company had completed a $500 million round that valued the company at $3.2 billion. At the time of his ouster, Mr. Kentley-Klay said in a statement that the board abruptly fired him and "chose a path of fear, optimizing for a little money in hand at the expense of profound progress for the universe." Mr. Kentley-Klay, an Australian designer, co-founded the company in 2014 with Jesse Levinson, who had made a name for himself in self-driving car development at Stanford University.