Improved sales and cost cuts helped automaker Honda Motor Co. shrug off lingering troubles from the Takata air bag recalls to log a nearly 19 percent improvement in its fiscal first-quarter profit. Tokyo-based Honda, which makes the Fit subcompact, Accord sedan and Asimo robot, on Tuesday reported a ¥207.3 billion profit for the April-June period, up from ¥174.6 billion the same period last year. Quarterly sales rose 7 percent to just over ¥3.7 trillion. Honda expects to sell 5.08 million vehicles for the fiscal year through March 2018, up from the nearly 5.03 million vehicles it sold during the last fiscal year. Honda raised its annual profit forecast to ¥545 billion from April's forecast of ¥530 billion. Its made a profit of ¥616.5 billion in the previous year.

A recent survey of Artificial Intelligence (AI) educators by Michael Wollowski, Peter Norvig and others (Wollowski et al., 2016) showed a stark difference of opinion about the definition of Artificial Intelligence. We invite you to participate in our survey to gather opinions on definitions of intelligence and Machine Intelligence from leading researchers. Understanding intelligence and how it may be recreated (and measured) is one of the major scientific challenges of our time. Our research shows that theories of intelligence and the goal of AI have been the source of much confusion both within the field and among the general public. This survey is completed anonymously but if you would like to be notified when the paper is available or have your definition of intelligence be considered for inclusion (with your name alongside) in our coming research paper, then there is also the opportunity to add your name and email address.

Lecture 1 of Tim Roughgarden's Algorithmic Game Theory class at Stanford (Autumn 2013) Class description: Topics at the interface of computer science and game theory such as: algorithmic mechanism design; combinatorial auctions; computation of Nash equilibria and relevant complexity theory; congestion and potential games; cost sharing; game theory and the Internet; matching markets; network formation; online learning algorithms; price of anarchy; prior-free auctions; selfish routing; sponsored search.

During the millions of pregnancies that occur in the United States every year, expectant moms learn oodles about their developing fetuses over months of gestation. But the placenta, a vital and temporary organ that shelters the fetus--delivering life-sustaining nutrients and oxygen, getting rid of toxic by-products and modulating the immune system to protect the pregnancy--largely remains a mystery. A team of Children's National Health System research scientists is beginning to provide insights about the poorly understood placenta. Using three-dimensional (3D) magnetic resonance imaging (MRI), the research team characterized the shape, volume, morphometry and texture of placentas during pregnancy and, using a novel framework, predicted with high accuracy which pregnancies would be complicated by fetal growth restriction (FGR). "When the placenta fails to carry out its essential duties, both the health of the mother and fetus can suffer and, in extreme cases, the fetus can die. Because there are few non-invasive tools that reliably assess the health of the placenta during pregnancy, unfortunately, placental disease may not be discovered until too late--after impaired fetal growth already has occurred," says Catherine Limperopoulos, Ph.D., co-director of research in the Division of Neonatology at Children's National Health System and senior author of the study published online July 22 in Journal of Magnetic Resonance Imaging.

WIRE)--Today's most successful brands already use data science to personalize experiences and address customer needs. Brands are increasingly building internal statistical models and algorithms to tailor experiences and yet most are not fully leveraging artificial intelligence (AI) technology. To help with this, Adobe (Nasdaq:ADBE) today announced it will open up its data science and algorithmic optimization capabilities in Adobe Target, the personalization engine of Adobe Marketing Cloud. Brands will be able to insert their own data models and algorithms into Adobe Target to deliver the best experience to customers. Additionally, the company announced new capabilities in Adobe Target powered by Adobe Sensei, its AI and machine learning framework, to further enhance customer recommendations and targeting precision, optimize experiences and automate the delivery of personalized offers.

Datameer recently announced SmartAI that integrates Big Data with machine learning models to provide better data insights. The goal of SmartAI is to be able to operationalize AI for Big Data i.e. run AI algorithms at scale on big data, to enable business analysts to plugin AI models into analytic workflows, and to be able to do all this in a secure and compliant manner. InfoQ caught up with Andrew Brust, senior director, market strategy and intelligence at Datameer, regarding the general role of Big Data in AI in general and SmartAI in particular. InfoQ: Datameer has hitherto been in the Big Data space. Why move to AI? Is Big Data passe?

Deep learning (DL) is a branch of machine learning that learns multiple levels of representations from data [1, 2]. DL has been successfully applied in pattern recognition and classification tasks such as image recognition and language processing. Recently, the application of DL to physics research is rapidly growing, such as in particle physics [3-7], nuclear physics [8], and condensed matter physics [9-14]. DL is shown to be very powerful in extracting pertinent features especially for complex nonlinear systems with high-order correlations that conventional techniques are unable to tackle. This suggests that it could be utilized to unveil hidden information from the highly implicit data of heavy-ion experiments.

Recent work in learning ontologies (hierarchical and partially-ordered structures) has leveraged the intrinsic geometry of spaces of learned representations to make predictions that automatically obey complex structural constraints. We explore two extensions of one such model, the order-embedding model for hierarchical relation learning, with an aim towards improved performance on text data for commonsense knowledge representation. Our first model jointly learns ordering relations and non-hierarchical knowledge in the form of raw text. Our second extension exploits the partial order structure of the training data to find long-distance triplet constraints among embeddings which are poorly enforced by the pairwise training procedure. We find that both incorporating free text and augmented training constraints improve over the original order-embedding model and other strong baselines.

The emergence of high-dimensional data, such as the gene expression values in microarray and the single nucleotide polymorphism data, brings challenges to many traditional statistical methods and theory. One important aspect of the high-dimensional data under the regression setting is that the number of covariates greatly exceeds the sample size. For example, in microarray data, the number of genes (p) is in the order of thousands whereas the sample size (n) is much less, usually less than 50. This is the so called "large-p, small-n" paradigm, which translates to a regime of asymptotics where p much faster than n. Inference in regression setting for large p, small n settings, have been recently developed. Sparsity assumption on the model signals has had a significant role in achieving optimal inference - Cai and Guo (2015); Javanmard and Montanari (2015); Cai and Guo (2016) found minimax results quantifying the direct effect of the size of the sparsity. In this article, we develop a test statistic that is able to quantify the simultaneous effect of a growing number of signals in a general high-dimensional linear model framework, allowing for a broad-ranging parameter structure.

We present a new topic model that generates documents by sampling a topic for one whole sentence at a time, and generating the words in the sentence using an RNN decoder that is conditioned on the topic of the sentence. We argue that this novel formalism will help us not only visualize and model the topical discourse structure in a document better, but also potentially lead to more interpretable topics since we can now illustrate topics by sampling representative sentences instead of bag of words or phrases. We present a variational auto-encoder approach for learning in which we use a factorized variational encoder that independently models the posterior over topical mixture vectors of documents using a feed-forward network, and the posterior over topic assignments to sentences using an RNN. Our preliminary experiments on two different datasets indicate early promise, but also expose many challenges that remain to be addressed.