In real-world scenarios, it is appealing to learn a model carrying out stochastic operations internally, known as stochastic computation graphs (SCGs), rather than learning a deterministic mapping. However, standard backpropagation is not applicable to SCGs. We attempt to address this issue from the angle of cost propagation, with local surrogate costs, called Q-functions, constructed and learned for each stochastic node in an SCG. Then, the SCG can be trained based on these surrogate costs using standard backpropagation. We propose the entire framework as a solution to generalize backpropagation for SCGs, which resembles an actor-critic architecture but based on a graph. For broad applicability, we study a variety of SCG structures from one cost to multiple costs. We utilize recent advances in reinforcement learning (RL) and variational Bayes (VB), such as off-policy critic learning and unbiased-and-low-variance gradient estimation, and review them in the context of SCGs. The generalized backpropagation extends transported learning signals beyond gradients between stochastic nodes while preserving the benefit of backpropagating gradients through deterministic nodes. Experimental suggestions and concerns are listed to help design and test any specific model using this framework.

Probabilistic representations, such as Bayesian and Markov networks, are fundamental to much of statistical machine learning. Thus, learning probabilistic representations directly from data is a deep challenge, the main computational bottleneck being inference that is intractable. Tractable learning is a powerful new paradigm that attempts to learn distributions that support efficient probabilistic querying. By leveraging local structure, representations such as sum-product networks (SPNs) can capture high tree-width models with many hidden layers, essentially a deep architecture, while still admitting a range of probabilistic queries to be computable in time polynomial in the network size. The leaf nodes in SPNs, from which more intricate mixtures are formed, are tractable univariate distributions, and so the literature has focused on Bernoulli and Gaussian random variables. This is clearly a restriction for handling mixed discrete-continuous data, especially if the continuous features are generated from non-parametric and non-Gaussian distribution families. In this work, we present a framework that systematically integrates SPN structure learning with weighted model integration, a recently introduced computational abstraction for performing inference in hybrid domains, by means of piecewise polynomial approximations of density functions of arbitrary shape. Our framework is instantiated by exploiting the notion of propositional abstractions, thus minimally interfering with the SPN structure learning module, and supports a powerful query interface for conditioning on interval constraints. Our empirical results show that our approach is effective, and allows a study of the trade off between the granularity of the learned model and its predictive power.

We propose Object-oriented Neural Programming (OONP), a framework for semantically parsing documents in specific domains. Basically, OONP reads a document and parses it into a predesigned object-oriented data structure (referred to as ontology in this paper) that reflects the domain-specific semantics of the document. An OONP parser models semantic parsing as a decision process: a neural net-based Reader sequentially goes through the document, and during the process it builds and updates an intermediate ontology to summarize its partial understanding of the text it covers. OONP supports a rich family of operations (both symbolic and differentiable) for composing the ontology, and a big variety of forms (both symbolic and differentiable) for representing the state and the document. An OONP parser can be trained with supervision of different forms and strength, including supervised learning (SL) , reinforcement learning (RL) and hybrid of the two. Our experiments on both synthetic and real-world document parsing tasks have shown that OONP can learn to handle fairly complicated ontology with training data of modest sizes.

Residual Networks make the very deep convolutional architecture works well, which use the residual unit to supplement the identity mappings. On the other hand, Squeeze-Excitation (SE) network propose an adaptively recalibrates channel-wise attention approach to model the relationship of feature maps from different convolutional channel. In this work, we propose the competitive SE mechanism for residual network, rescaling value for each channel in this structure will be determined by residual and identity mappings jointly, this design enables us to expand the meaning of channel relationship modeling in residual blocks: the modeling of competition between residual and identity mappings make identity flow can controll the complement of residual feature maps for itself. Further, we design a novel pair-view competitive SE block to shrink the consumption and re-image the global characterizations of intermediate convolutional channels. We carry out experiments on datasets: CIFAR, SVHN, ImageNet, the proposed method can be compared with the state-of-the-art results.

AI is no longer the stuff of science fiction – it's impacting patients today It's difficult to open a newspaper nowadays without seeing an article about artificial intelligence. But one thing you cannot escape is that AI is here now and it's only going to become more pervasive. While fear of an unknown technology is understandable, in many ways it does a disservice to the incredible impact that AI is already having on the world around us. In the healthcare space alone, it is offering ways to fundamentally rethink clinical practice, speeding up diagnosis, driving patient support programs and aiding drug discovery. In only five years, more than 200 venture capital and private equity deals to fund research into the use of AI in healthcare have been signed.

Your Shopping Cart is empty. How companies are using artificial intelligence in their business operations. A 5-Part Process for Using Technology to Improve Your Talent Management Talent management Digital Article "Talent tech" can change how firms hire, evaluate, and develop employees. We Need Transparency in Algorithms, But Too Much Can Backfire Technology Digital Article Consumers should have some idea how machines make decisions. Most of AI's Business Uses Will Be in Two Areas Technology Research Supply chain and sales and marketing are the first big opportunities.

They can unload the dishwasher, deliver packages to your home and open doors. Their thin, metallic legs are able to traverse a steep flight of stairs -- or crawl straight into your worst nightmares. Now Boston Dynamics' awkward, four-legged, doglike robot, SpotMini, is evolving from a YouTube sensation to a purchasable pet of sorts, according to the company's founder, Marc Raibert. Raibert told an audience last month at the CeBIT computer expo in Hanover, Germany, that his company is already testing SpotMini with potential customers from four separate industries: security, delivery, construction and home assistance. His presentation at the expo was reported by Inverse.

A man is reflected on an electronic board showing the graphs of Japan's Nikkei share average. Commodities big data is a precious resource, but how can companies make the most of their unstructured information to secure a knowledge advantage? Commodity markets are shifting at an unprecedented pace, and the volume of data that is their essence has grown exponentially over the course of the past few years. Digital transformation is happening at different speeds in different commodities markets, but there is a growing awareness that failure to embrace digital is no longer an option. Back in the boom times, it was not imperative for companies to deep dive into data.

On the heels of United States President Donald Trump's historic de-nuclearization summit with North Korean leader Kim Jong-un, non-proliferation is once again a timely topic. Since the dawn of the nuclear age, keeping tabs on who has military-grade nuclear capabilities and materials has been a vital – and difficult – task. Thankfully, it's also one that may be getting easier, thanks to leaps forward in fields like data analysis, machine learning and artificial intelligence. Last month, Thomson Reuters Labs was invited to present at a workshop called "Applications of Innovative Tools and Technologies for Nonproliferation and Disarmament" held in Krems, Austria, for diplomats representing their countries at the International Atomic Energy Agency (IAEA) and other international organizations. The diplomatic workshop was preceded by a day-long session for technical participants at the Vienna Center for Disarmament and Non-Proliferation.

The chief executive officer of government contractor Booz Allen Hamilton Inc. warned that the U.S. has only a small advantage over China in the rising field of artificial intelligence and is at risk of falling behind without a "national strategy." "It's not the 50-year edge that we have in building aircraft carriers," Horacio Rozanski said Thursday in a meeting with Bloomberg editors and reporters in Washington. Chinese President Xi Jinping has made a 10-fold increase in AI output a national priority as the world's second-largest economy seeks to dominate the industry by 2030. The U.S. has little formal AI strategy at the federal level, although resources and government projects have accelerated in the last year or so, Rozanski said. "The investments are being driven at this point by the strategies of the different parts of the government, as opposed to collective strategy," he said.