Integrating model-free and model-based approaches in reinforcement learning has the potential to achieve the high performance of model-free algorithms with low sample complexity. However, this is difficult because an imperfect dynamics model can degrade the performance of the learning algorithm, and in sufficiently complex environments, the dynamics model will almost always be imperfect. As a result, a key challenge is to combine model-based approaches with model-free learning in such a way that errors in the model do not degrade performance. We propose stochastic ensemble value expansion (STEVE), a novel model-based technique that addresses this issue. By dynamically interpolating between model rollouts of various horizon lengths for each individual example, STEVE ensures that the model is only utilized when doing so does not introduce significant errors. Our approach outperforms model-free baselines on challenging continuous control benchmarks with an order-of-magnitude increase in sample efficiency, and in contrast to previous model-based approaches, performance does not degrade in complex environments.

We consider doing Bayesian inference by minimizing the KL divergence on the 2-Wasserstein space $\mathcal{P}_2$. By exploring the Riemannian structure of $\mathcal{P}_2$, we develop two inference methods by simulating the gradient flow on $\mathcal{P}_2$ via updating particles, and an acceleration method that speeds up all such particle-simulation-based inference methods. Moreover we analyze the approximation flexibility of such methods, and conceive a novel bandwidth selection method for the kernel that they use. We note that $\mathcal{P}_2$ is quite abstract and general so that our methods can make closer approximation, while it still has a rich structure that enables practical implementation. Experiments show the effectiveness of the two proposed methods and the improvement of convergence by the acceleration method.

Deep neural networks have shown promising results for various clinical prediction tasks such as diagnosis, mortality prediction, predicting duration of stay in hospital, etc. However, training deep networks -- such as those based on Recurrent Neural Networks (RNNs) -- requires large labeled data, high computational resources, and significant hyperparameter tuning effort. In this work, we investigate as to what extent can transfer learning address these issues when using deep RNNs to model multivariate clinical time series. We consider transferring the knowledge captured in an RNN trained on several source tasks simultaneously using a large labeled dataset to build the model for a target task with limited labeled data. An RNN pre-trained on several tasks provides generic features, which are then used to build simpler linear models for new target tasks without training task-specific RNNs. For evaluation, we train a deep RNN to identify several patient phenotypes on time series from MIMIC-III database, and then use the features extracted using that RNN to build classifiers for identifying previously unseen phenotypes, and also for a seemingly unrelated task of in-hospital mortality. We demonstrate that (i) models trained on features extracted using pre-trained RNN outperform or, in the worst case, perform as well as task-specific RNNs; (ii) the models using features from pre-trained models are more robust to the size of labeled data than task-specific RNNs; and (iii) features extracted using pre-trained RNN are generic enough and perform better than typical statistical hand-crafted features.

The folks who brought us the marshmallow test have some unlikely news: children today have more self-control than ever. That conclusion is based on more than 50 years of results from the iconic test, which allows a preschooler to eat one treat immediately or two if she can wait 10 minutes. The effort at delayed gratification is vastly funny but the results were found to have serious implications for children's future success. Led by psychologist Walter Mischel, who created the experiment -- one of the most famous in developmental psychology -- a research team found that children tested between 2002-2012 held out for two minutes longer on average than the original test-takers in the 1960s, and one minute longer than participants in the 1980s. A 4-year-old in the earliest group waited as long as a child between 2 ½ and 3 in the most recent tests, and 4-year-old test-takers in the 1980s waited as long as a child who was 3 ½ in the 2000s.

Lumboon Simakajornboon, ABB's local business unit manager for robotics and motion, said the Thai automobile industry has pioneered the use of robotic and automation in its production process, but now major players in food and beverage as well as other industries, such as CP Group, Betrago, Singha, Thai Beverage Group, and Red Bull are also investing heavily in robotic and automation systems. Besides the production process, robotics and automation are used in packaging and moving pallets of finished products onto transport trucks in a bid to increase efficiency and speed for the delivery of products to consumers, he said, adding that major retail and supercenter chains such as Lotus also have installed an automated system for inventory management and logistics. For food and beverage companies, traceability of the raw materials and other ingredients used in the production process has become a new requirement to ensure product safety and the ability to trace back when there are quality and other problems. To facilitate traceablity and other new requirements, machines used factories have become connected to send and recieve data while analytics is used for real-time processing via cloud-based computing facilities. Lumboon said the Thai food and beverage industry is a key target for robotics and automation due to its large footprint with as many as 10,000 producers, including many small and medium-sized enterprises.

I can't run a starship with twenty crew. KIRK: And what am I supposed to do? WESLEY: You've got a great job, Jim. All you have to do is sit back and let the machine do the work. One clear message from the presidential election is that the dream of good factory jobs still resonates in America's rust belt. Despite the push for students to pursue STEM careers or move into the service sector, Americans still want to make stuff.

When a CIA-backed venture capital fund took an interest in Rana el Kaliouby's face-scanning technology for detecting emotions, the computer scientist and her colleagues did some soul-searching -- and then turned down the money. "We're not interested in applications where you're spying on people," said el Kaliouby, the CEO and co-founder of the Boston startup Affectiva. The company has trained its artificial intelligence systems to recognize if individuals are happy or sad, tired or angry, using a photographic repository of more than 6 million faces. Recent advances in AI-powered computer vision have accelerated the race for self-driving cars and powered the increasingly sophisticated photo-tagging features found on Facebook and Google. But as these prying AI "eyes" find new applications in store checkout lines, police body cameras and war zones, the tech companies developing them are struggling to balance business opportunities with difficult moral decisions that could turn off customers or their own workers.

"AI could play a big role in supporting prevention, diagnosis, treatment plans, medication management, precision medicine and drug creation" – Bruce Lang Artificial intelligence, the capability of a machine to reflect human behavior in an intelligent manor, has the opportunity to revolutionize healthcare. These algorithms and deep learning processes evaluate and study data, utilizing their findings to apply to problems. When the data is inputed into the algorithm, the information is more understood throughout time and the AI is able to provide exponentially better feedback. Artificial intelligence can, for example, exam a patient medical history and predict the likelihood of that individual having a certain disease. Such uses are for hospital management, drug discovery, lifestyle monitoring, patient risk analysis, diagnostics, and more.

The leading Chinese-language browser is searching for a new identity--no longer plodding internet-service firm, but artificial-intelligence dynamo. Its ambition will be on display today in Beijing at Baidu Create 2018, the second year of an annual two-day shindig for AI developers. Awaiting attendees are the latest versions of Apollo, an open autonomous-driving platform, and DuerOS, Baidu's freshly minted AI assistant. Both were spearheaded by Lu Qi, a former Microsoft executive hired by Baidu as its star AI expert in early 2017--until he left in May. The firm's shares duly tumbled by 9.5%, the most in three years, and Credit Suisse downgraded them; investors were reminded of the premature departures of Andrew Ng, Baidu's former chief scientist, and Wang Jing, the head of its self-driving unit. Few expect Mr Lu's exit to force a drastic rethink at Baidu, but it could now take longer for its AI business to drive a profit.

Planck Re, a startup that wants to simplify insurance underwriting with artificial intelligence, announced today that it has raised a $12 million Series A. The funding was led by Arbor Ventures, with participation from Viola FinTech and Eight Roads. Co-founder and CEO Elad Tsur tells TechCrunch that the capital will be used to expand Planck Re's product line into more segments, including retail, contractors, IT and manufacturing, and grow its research and development team in Israel and North American sales team. The Tel Aviv and New York-based startup plans to focus first on its business in the United States, where it has already launched pilot programs with several insurance carriers. Tsur says that Planck Re's clients generally use it to help underwrite insurance for small to medium-sized businesses, including business owner policies, which cover property and liability risks, and workers' compensation. Founded in 2016 by Tsur, Amir Cohen and David Schapiro, Planck Re poses its technology as a more efficient and accurate alternative to the lengthy risk assessment questionnaire insurers ask clients to fill out.