One big change is support for running trained machine learning models as part of Windows applications, taking advantage of local GPUs to accelerate machine learning applications. Building a machine learning application can be a complex process. Training a model can require a lot of data, and a considerable amount of processing power. That's fine if you've got access to a cloud platform and lots of bandwidth, but what if you want to take an existing model from GitHub and run it on a PC? Trained machine learning models are an ideal tool for bringing the benefits of neural networks and deep learning to your applications. All you should need to do is hook up the appropriate interfaces, and they should run as part of your code.

Chinese retail giant Alibaba claims to have created an artificial intelligence capable of beating the Turing test - and writing 20,000 lines of advertisement copy in a single second. Using deep learning and natural language processing technology, the system has been taught to produce copy by being shown millions of examples of human-written descriptions of products sold online. Produced by Alimama, a marketing division of Alibaba, the AI is currently being offered to vendors on Tmall and Taobao, Alibaba's e-commerce platforms, as a speedy alternative to writing their own copy. When listing an item for sale, vendors can click a button marked'Produce Smart Copy' and instantly see multiple copy ideas. Although Alibaba has not provided any proof of its AI passing the Turing test, this is a sizeable claim.

There are lots of education options available online, provided you're a self-starter with the discipline to do a lot of coursework on your own. For example, Microsoft's AI School offers a variety of lessons for developers in everything from text analytics and object recognition to custom neural-network models. The content is angled toward data scientists and developers, and heavily emphasizes the use of Microsoft products (of course) in addition to "universal" A.I. skills. It's also free, although those who want Verified Certificates will need to pay a fee. Microsoft, of course, is far from your only option when it comes to learning about A.I. online, particularly with regard to beginner-level material.

Abdominal ultrasound tests for organ abnormalities haven't changed much in the past decade, with a doctor moving a wand over the patient's abdomen to gaze at blurry images. But the process could get accelerated by a thousand times with improved accuracy, based on deep learning work by U.S. researchers. Tests typically take around half an hour. That may not seem like a long time, but given that hospitals do thousands of these scans every year, they end up spending massive amounts of time providing this procedure when they could be seeing more patients. These examinations, done to diagnose abnormalities in various internal organs including the kidneys, liver or gallbladder, require substantial effort from practitioners who must find the right angle for ultrasound imaging, annotate these views in text and record relevant measurements.

There is a new and apparently better way to fix grainy photos, and it uses AI. Artificial Intelligence, one must say, coupled with lots of computing power, but that's growing day after day. Artificial Intelligence and machine learning will never cease to surprise us, apparently. Each new day, there's some new announcement, covering different fields: AI can render 3D hair in real time, smell illnesses in human breath, assess infrastructure quality in Africa or help transform audio into music playing avatars. Now it can also help photographers get rid of noise in their photos.

We are big fan of cloud-based machine learning and deep learning, and AI in general. After all, you can't be a geek without imagining having a conversation with an artificially intelligent being that can answer questions and carry out your bidding! That's said, we are also seeing cloud-based machine learning and deep learning misapplied over and over again. All have easy fixes for the most part, and certainly cloud-based machine learning is here to stay. But use it wisely and appropriately.

"Computer vision" has a decidedly sci-fi ring, so it's no surprise the idea of smart devices becoming all-seeing tools has captured people's imaginations. Yet the true potential of the tech is held back by a widespread misunderstanding of what great applications look like and what they can achieve. Where did computer vision come from? It's the product of a proliferation of cheap, high-quality cameras, which has expanded the scope for imagery captured in public, private, and commercial domains. At the same time, advances in machine learning and deep learning technology are allowing us to transform those images into digital signals that support a wide range of actions.

Most policy search algorithms require thousands of training episodes to find an effective policy, which is often infeasible with a physical robot. This survey article focuses on the extreme other end of the spectrum: how can a robot adapt with only a handful of trials (a dozen) and a few minutes? By analogy with the word "big-data", we refer to this challenge as "micro-data reinforcement learning". We show that a first strategy is to leverage prior knowledge on the policy structure (e.g., dynamic movement primitives), on the policy parameters (e.g., demonstrations), or on the dynamics (e.g., simulators). A second strategy is to create data-driven surrogate models of the expected reward (e.g., Bayesian optimization) or the dynamical model (e.g., model-based policy search), so that the policy optimizer queries the model instead of the real system. Overall, all successful micro-data algorithms combine these two strategies by varying the kind of model and prior knowledge. The current scientific challenges essentially revolve around scaling up to complex robots (e.g., humanoids), designing generic priors, and optimizing the computing time.

Deep learning usually requires big data, with respect to both volume and variety. However, most remote sensing applications only have limited training data, of which a small subset is labeled. Herein, we review three state-of-the-art approaches in deep learning to combat this challenge. The first topic is transfer learning, in which some aspects of one domain, e.g., features, are transferred to another domain. The next is unsupervised learning, e.g., autoencoders, which operate on unlabeled data. The last is generative adversarial networks, which can generate realistic looking data that can fool the likes of both a deep learning network and human. The aim of this article is to raise awareness of this dilemma, to direct the reader to existing work and to highlight current gaps that need solving.

Recognizing apparel attributes has recently drawn great interest in the computer vision community. Methods based on various deep neural networks have been proposed for image classification, which could be applied to apparel attributes recognition. An interesting problem raised is how to ensemble these methods to further improve the accuracy. In this paper, we propose a two-layer mixture framework for ensemble different networks. In the first layer of this framework, two types of ensemble learning methods, bagging and boosting, are separately applied. Different from traditional methods, our bagging process makes use of the whole training set, not random subsets, to train each model in the ensemble, where several differentiated deep networks are used to promote model variance. To avoid the bias of small-scale samples, the second layer only adopts bagging to mix the results obtained with bagging and boosting in the first layer. Experimental results demonstrate that the proposed mixture framework outperforms any individual network model or either independent ensemble method in apparel attributes classification.