In this paper, we present an initial attempt to learn evolution PDEs from data. Inspired by the latest development of neural network designs in deep learning, we propose a new feed-forward deep network, called PDE-Net, to fulfill two objectives at the same time: to accurately predict dynamics of complex systems and to uncover the underlying hidden PDE models. The basic idea of the proposed PDE-Net is to learn differential operators by learning convolution kernels (filters), and apply neural networks or other machine learning methods to approximate the unknown nonlinear responses. Comparing with existing approaches, which either assume the form of the nonlinear response is known or fix certain finite difference approximations of differential operators, our approach has the most flexibility by learning both differential operators and the nonlinear responses. A special feature of the proposed PDE-Net is that all filters are properly constrained, which enables us to easily identify the governing PDE models while still maintaining the expressive and predictive power of the network. These constrains are carefully designed by fully exploiting the relation between the orders of differential operators and the orders of sum rules of filters (an important concept originated from wavelet theory). We also discuss relations of the PDE-Net with some existing networks in computer vision such as Network-In-Network (NIN) and Residual Neural Network (ResNet). Numerical experiments show that the PDE-Net has the potential to uncover the hidden PDE of the observed dynamics, and predict the dynamical behavior for a relatively long time, even in a noisy environment.

Iberian winemakers have tended vineyards by hand since before the Roman Empire. Today, traditional winemaking techniques still hold sway in the third largest wine producing nation in the world. But 2016 could be the year it all changes. Even though less than 10 percent of Spanish wineries use advanced technologies now commonly seen in places like Australia and the United States, "there are many that have tried them this year," says Fran Garcia Ruiz, director of agricultural data company AgroMapping. Spanish wineries, which have more than 2.9 million acres of vineyards, have lagged behind counterparts elsewhere in adopting new technology.

Linear Dynamical Systems (LDSs) are fundamental tools for modeling spatio-temporal data in various disciplines. Though rich in modeling, analyzing LDSs is not free of difficulty, mainly because LDSs do not comply with Euclidean geometry and hence conventional learning techniques can not be applied directly. In this paper, we propose an efficient projected gradient descent method to minimize a general form of a loss function and demonstrate how clustering and sparse coding with LDSs can be solved by the proposed method efficiently. To this end, we first derive a novel canonical form for representing the parameters of an LDS, and then show how gradient-descent updates through the projection on the space of LDSs can be achieved dexterously. In contrast to previous studies, our solution avoids any approximation in LDS modeling or during the optimization process. Extensive experiments reveal the superior performance of the proposed method in terms of the convergence and classification accuracy over state-of-the-art techniques.

LUIS is now generally available in the Australia East, Brazil South, West US 2, South Central US, East US, East Asia, and North Europe regions, in addition to the current availability in the East US 2, West Central US, West US, West Europe, and Southeast Asia regions. General availability (GA) pricing will begin on February 1, 2018. Usage prior to February 1, 2018, will be billed at preview rates.

Forgetting is an operation on knowledge bases that has been addressed in different areas of Knowledge Representation and with respect to different formalisms, including classical propositional and first-order logic, modal logics, logic programming, and description logics. Definitions of forgetting have been expressed in terms of manipulation of formulas, sets of postulates, isomorphisms between models, bisimulations, second-order quantification, elementary equivalence, and others. In this paper, forgetting is regarded as an abstract belief change operator, independent of the underlying logic. The central thesis is that forgetting amounts to a reduction in the language, specifically the signature, of a logic. The main definition is simple: the result of forgetting a portion of a signature in a theory is given by the set of logical consequences of this theory over the reduced language. This definition offers several advantages. Foremost, it provides a uniform approach to forgetting, with a definition that is applicable to any logic with a well-defined consequence relation. Hence it generalises a disparate set of logic-specific definitions with a general, high-level definition. Results obtained in this approach are thus applicable to all subsumed formal systems, and many results are obtained much more straightforwardly. This view also leads to insights with respect to specific logics: for example, forgetting in first-order logic is somewhat different from the accepted approach. Moreover, the approach clarifies the relation between forgetting and related operations, including belief contraction.

Google's AI system assisted researchers in New Zealand in identifying calls of native birds -- Kakariki and Hihi -- using acoustic sensors after sifting through 15,000 hours of audio captured in and around Wellington. The Google Brain team, a core group focused on deep learning, used a trained Tensorflow model to label spectrograms and validate results to classify bird songs in real time. Tensorflow is an open source software for machine learning (ML) developed by the Google Brain team that was launched in 2015. Since then, Google has been running ML on different data sets -- from tracking seacows to diagnosing diabetic retinopathy and other health challenges. Linne Ha, director of Google Research and Machine Intelligence, shared updates on Google's ambitious Project Unison that's attempting to create text-tospeech (TTS) voices for lowresourced languages. There are 6,000 languages globally and 400 of them have over a million speakers.

EXCLUSIVE – Blockchain in cross-border payments will be big in 2018. At least that's what IBM's Rajesh Venkatraman, director of worldwide payments solutions sales at IBM, thinks. "Blockchain technology has many applications, but I believe cross-border payments is especially going to see tremendous traction," he told Bank Innovation. Perhaps that's also because IBM has started numerous pilots in this area. Earlier this fall, it announced a collaboration with startup Stellar, which uses blockchain technology to connect fiat currencies to enable instant international transfers.

Was this the legal sector's "Kodak moment"? The event that signalled the beginning of the end: "The people of Darwin can just about take the law into their own hands, with a new legal firm going lawyer-free," ABC News reported recently. "With a few clicks of a button, a client can enter their details and will then be asked a few simple questions by Ailira, before the robot generates a fully certified will, using the Ailira system." Right now, Australia has only a handful of businesses that have successfully integrated artificial intelligence into their day-to-day operations. But each month we are witnessing advancements and seeing early adopters reap the benefits.

Decision tree classifiers are a widely used tool in data stream mining. The use of confidence intervals to estimate the gain associated with each split leads to very effective methods, like the popular Hoeffding tree algorithm. From a statistical viewpoint, the analysis of decision tree classifiers in a streaming setting requires knowing when enough new information has been collected to justify splitting a leaf. Although some of the issues in the statistical analysis of Hoeffding trees have been already clarified, a general and rigorous study of confidence intervals for splitting criteria is missing. We fill this gap by deriving accurate confidence intervals to estimate the splitting gain in decision tree learning with respect to three criteria: entropy, Gini index, and a third index proposed by Kearns and Mansour. We also extend our confidence analysis to a selective sampling setting, in which the decision tree learner adaptively decides which labels to query in the stream. We provide theoretical guarantees bounding the probability that the decision tree learned via our selective sampling strategy classifies suboptimally the next example in the stream. Experiments on real and synthetic data in a streaming setting show that our trees are indeed more accurate than trees with the same number of leaves generated by state-of-the-art techniques. In addition to that, our active learning module empirically uses fewer labels without significantly hurting the performance.

Could robots change the way we think? While that might seem the stuff of dark science fiction, New Zealand artificial intelligence (AI) experts say there's real fear that computer algoritms could hijack our language, and ultimately influence our views on products or politics. "I would compare the situation with the subliminal advertising that was outlawed in the 1970s," said Associate Professor Christoph Bartneck, of Canterbury University's Human Interface Technology Laboratory, or HIT Lab. "We are in a danger of repeated the exact same issue with the use of our language." Bartneck has been working in the area with colleague Jurgen Brandstetter and other experts at the New Zealand Institute of Language Brain and Behaviour and Northwestern University in the US.