This paper proposes a tentative and original survey of meeting points between Knowledge Representation and Reasoning (KRR) and Machine Learning (ML), two areas which have been developing quite separately in the last three decades. Some common concerns are identified and discussed such as the types of used representation, the roles of knowledge and data, the lack or the excess of information, or the need for explanations and causal understanding. Then some methodologies combining reasoning and learning are reviewed (such as inductive logic programming, neuro-symbolic reasoning, formal concept analysis, rule-based representations and ML, uncertainty in ML, or case-based reasoning and analogical reasoning), before discussing examples of synergies between KRR and ML (including topics such as belief functions on regression, EM algorithm versus revision, the semantic description of vector representations, the combination of deep learning with high level inference, knowledge graph completion, declarative frameworks for data mining, or preferences and recommendation). This paper is the first step of a work in progress aiming at a better mutual understanding of research in KRR and ML, and how they could cooperate.

Artificial intelligence can perform feats that seem like sorcery. AI can drive cars and fly drones. It can compose original music, write poetry that isn't too awful, and design recipes that do sound awful (blueberry and spinach pizza, anyone?). AI can do some things better than humans: lip reading, diagnosing diseases such as pneumonia and some cancers, transcribing speech, and playing Jeopardy!, Go, Texas Hold'em, and a variety of video games. AI software can even learn to make its own AI software.

Technology develops differently in Japan. While US tech giants have been grabbing artificial intelligence headlines, a business AI sector has been quietly maturing in Japan, and it is now making inroads into America. Today we sit down again with Miku Hirano, CEO of Cinnamon, and we talk about how exactly this happened. Interestingly, Cinnamon did not start out as an AI company. In fact, when Miku first came on the show, the company had just launched an innovative video-sharing service. Today, we talk about what lead to the pivot to AI and why even a great idea and a great team is no guarantee of success. We also talk about some of the changing attitudes towards startups and women in Japan, the kinds of business practices AI will never change, and Miku give some practical advice for startups going into foreign markets. It's a great discussion, and I think you will really enjoy it. Welcome to Disrupting Japan, straight talk from Japan's most successful entrepreneurs. Today, we're going to sit down and talk about artificial intelligence with Miku Hirano of Cinnamon. Now, Cinnamon is actually a great example of a successful Japanese startup pivot. When we first sat down with Miku four years ago, she had an innovative micro-video sharing company called Tuya and really, you should go back and listen to that episode. I've put a link on the show notes and it was really a good one.

Nearly 600 years after Chinese monks advanced the spread of knowledge with block printing, there was the Gutenberg Press, changing the dissemination of knowledge forever. And now, nearly 600 years later, Artificial Intelligence is poised to do the same. For many, Artificial Intelligence is clouded in mystery, and bound to big screen killer bots that ultimately decide that humans must be terminated. From HAL 9000, to Skynet, to I, Robot, and back to Skynet again, AI has been popularly framed in how it can hurt humanity, rather than how it can help. Artificial Intelligence aims to train machines to perform tasks with the hallmarks of human intelligence: inference, speech recognition, visual perception, planning, learning, and language comprehension.

I came to a shocking conclusion while writing about diversity for my book on machine learning: diversity in Artificial Intelligence is not your problem, it's hers. I mean, of course, that the problem is with the English pronoun "hers". There is a bias against "hers" in most major AI systems today, and the source of the bias is the perfect metaphor for bias in AI more broadly. Like you might remember from high school, "hers" is a pronoun. Each word in a sentence belongs to one of a small number of categories: nouns, pronouns, adjectives, verbs, adverbs, etc. One common building block in many AI applications is to identify the right category in raw text. Today, "hers" is not recognized as a pronoun by the most widely used technologies for Natural Language Processing (NLP), including (alphabetically) Amazon Comprehend, Google Natural Language API, and the Stanford Parser. The video shows that in the sentence "the car is hers", Amazon and Google classify "hers" as a noun and the Stanford parser classifies "hers" as an adjective. They don't make the same mistake with the sentence "the car is his", correctly identifying "his" as a pronoun.

A new informative report on the global Artificial Intelligence for Edge Devices Market titled as, Artificial Intelligence for Edge Devices has recently published by Contrive Market Research to its humongous database which helps to shape the future of the businesses by making well-informed business decisions. It offers a comprehensive analysis of various business aspects such as global market trends, recent technological advancements, market shares, size, and new innovations. Furthermore, this analytical data has been compiled through data exploratory techniques such as primary and secondary research. Moreover, an expert team of researchers throws light on various static as well as dynamic aspects of the global Artificial Intelligence for Edge Devices Market. Different leading key players have been profiled to get better insights into the businesses. It offers detailed elaboration on different top-level industries which are functioning in global regions.

Speech-driven facial animation involves using a speech signal to generate realistic videos of talking faces. Recent deep learning approaches to facial synthesis rely on extracting low-dimensional representations and concatenating them, followed by a decoding step of the concatenated vector. This accounts for only first-order interactions of the features and ignores higher-order interactions. In this paper we propose a polynomial fusion layer that models the joint representation of the encodings by a higher-order polynomial, with the parameters modelled by a tensor decomposition. We demonstrate the the suitability of this approach through experiments on generated videos evaluated on a range of metrics on video quality, audiovisual synchronisation and generation of blinks.

A dry decade in the Navajo Nation has killed vegetation, dessicated soils, and released once-stable sand into the wind. This sand now covers one-third of the Nation's land, threatening roads, gardens and hundreds of homes. Many arid regions have similar problems: global warming has increased dune movement across farmland in Namibia and Angola, and the southwestern US. Current dune models, unfortunately, do not scale well enough to provide useful forecasts for the $\sim$5\% of land surfaces covered by mobile sand. We test the ability of two deep learning algorithms, a GAN and a CNN, to model the motion of sand dunes. The models are trained on simulated data from community-standard cellular automaton model of sand dunes. Preliminary results show the GAN producing reasonable forward predictions of dune migration at ten million times the speed of the existing model.

We study the problem of off-policy evaluation (OPE) in Reinforcement Learning (RL), where the aim is to estimate the performance of a new policy given historical data that may have been generated by a different policy, or policies. In particular, we introduce a novel doubly-robust estimator for the OPE problem in RL, based on the Targeted Maximum Likelihood Estimation principle from the statistical causal inference literature. We also introduce several variance reduction techniques that lead to impressive performance gains in off-policy evaluation. We show empirically that our estimator uniformly wins over existing off-policy evaluation methods across multiple RL environments and various levels of model misspecification. Finally, we further the existing theoretical analysis of estimators for the RL off-policy estimation problem by showing their $O_P(1/\sqrt{n})$ rate of convergence and characterizing their asymptotic distribution.

It is the present-day darling of the tech world. The current renaissance of Artificial Intelligence (AI) with its sister discipline Machine Learning (ML) has led every IT firm worth its salt to engineer some form of AI onto its platform, into its toolsets and throughout its software applications. IBM CEO Ginni Rometty has already proclaimed that AI will change 100 percent of jobs over the next decade. And yes, she does mean everybody's job from yours to mine and onward to the role of grain farmers in Egypt, pastry chefs in Paris and dog walkers in Oregon i.e. every job. We will now be able to help direct all workers' actions and behavior with a new degree of intelligence that comes from predictive analytics, all stemming from the AI engines we will now increasingly depend upon.