Starting today, Google will rely more heavily on artificial intelligence when it translates language. The new method, called Google Machine Neural Translation, cuts down errors by 80 percent compared to its current algorithm, and is nearly indistinguishable from human translation on standardized tests, the company said. It's a radical change from how Google translates now, which is called phrase-based translation, and is standard for the industry. Under this method, an algorithm cuts up a sentence, like one entered by a high-schooler trying to game their homework, and attempts to match words or phrases to a large dictionary. The new method takes that same large dictionary and uses it to train two neural networks.

The deep learning software driving the modern artificial intelligence revolution has mostly run on fairly standard computer hardware. Some tech giants such as Google and Intel have focused some of their considerable resources on creating more specialized computer chips designed for deep learning. But IBM has taken a more unusual approach: It is testing its brain-inspired TrueNorth computer chip as a hardware platform for deep learning. Deep learning's powerful capabilities rely on algorithms called convolutional neural networks that consist of layers of nodes (also known as neurons). Such neural networks can filter huge amounts of data through their "deep" layers to become better at, say, automatically recognizing individual human faces or understanding different languages. These are the types of capabilities that already empower online services offered by the likes of Google, Facebook, Amazon and Microsoft.

Most people think Photoshop is a magical tool that can change reality, but it does require a skilled artist for decent results. Using neural networks, however, University of Edinburgh researcher Andrew Brock has built an uncanny image editing app that can transform someone's entire hairstyle with just the stroke of a brush. To use the Neural Photo Editor, you select the "contextual paintbrush," choose a color and start mousing the part you want to change. The system can recognize if you're on the model's hair, for instance, and intelligently fill in the area, changing the color to match your brush. In another example, the user paints over the subject's mouth with a white brush to make their smile bigger.

Big data is in many ways still a wild frontier, requiring wily smarts and road-tested persistence on the part of those hoping to find insight in all the petabytes. On Tuesday, IBM announced a new platform it hopes will make things easier. Dubbed Project DataWorks, the new cloud-based platform is the first to integrate all types of data and bring AI to the table for analytics, IBM said. Project DataWorks is available on IBM's Bluemix cloud platform and aims to foster collaboration among the many types of people who need to work with data. Tapping technologies including Apache Spark, IBM Watson Analytics and the IBM Data Science Experience launched in June, the new offering is designed to give users self-service access to data and models while ensuring governance and rapid-iteration capabilities. Project DataWorks can ingest data faster than any other data platform, from 50 to hundreds of Gbps, deriving from sources including enterprise databases, the internet of things (IoT) and social media, according to IBM.

But I am even more excited about the train your own model feature. We believe that the same AI technology that gives big tech companies a competitive edge should be available to developers or businesses of any size or budget. That's why we built our new Custom Training and Visual Search products – to make it easy, quick, and inexpensive for developers and businesses to innovate with AI, go to market faster, and build better user experiences. Machine learning requires large data sets and skilled engineers to build the technology that can derive "intelligence" from data. And so without machine learning as a service from companies like Clarifai, the largest tech companies will have a structural advantage over small developers.

Neural Photo Editor is an experimental piece of retouching software from researchers at the University of Edinburgh that uses neural networks to act like Photoshop on steroids. Thanks to machine learning, it can intuitively interpret how a user intends to retouch a photo based on a "contextual paintbrush." A single brush can change hair color, fill in bald spots, or add a toothy grin. The process couldn't be simpler: Users select a color for their paintbrush and the system analyzes that color in context with the image in order to produce an intelligent output. Painting over a subject's mouth with a white brush, for example, can make a smile larger, while painting with a dark color on a forehead can add bangs.

The question of which model type to apply to a Machine Learning task can be a daunting one given the immense number of algorithms available in the literature. It can be difficult to compare the relative merits of two methods, as one can outperform the other in a certain class of problems while consistently coming in behind for another class. In this post, the last one of our series of posts about Logistic Regression, we'll explore the differences between Decision Trees and Logistic Regression for classification problems, and try to highlight scenarios where one might be recommended over the other. Logistic Regression and trees differ in the way that they generate decision boundaries i.e. the lines that are drawn to separate different classes. To illustrate this difference, let's look at the results of the two model types on the following 2-class problem: Decision Trees bisect the space into smaller and smaller regions, whereas Logistic Regression fits a single line to divide the space exactly into two.

Francisco is the Founder and CEO of cortical.io, Francisco's medical background in genetics combined with over two decade's of experience in Information Technology, inspired him to create a groundbreaking technology, called Semantic Folding, which is based on the latest findings on the way the human neocortex processes information. Francisco founded Matrixware Information Services, a company that developed the first standardized database of patents. Francisco also initiated the Information Retrieval Facility, a non-profit research institute, with the goal to bridge the gap between science and industry in the information retrieval domain. Let me introduce you to Francisco Webber, Founder and CEO of cortical.io.

Our third regional GPU Technology Conference in as many weeks reached another packed house today, as NVIDIA co-founder and CEO Jen-Hsun Huang unveiled technology that will accelerate the deep learning revolution. "GPU computing is at the beginning of something very, very important, a brand new revolution, what people call the AI revolution, the beginning of the fourth industrial revolution," Huang told a crowd of 1,600 scientists, engineers, entrepreneurs and press, gathered at Amsterdam's gleaming waterfront music hall. "However you describe it, we think something really big is around the corner." In the latest stop in a tour that will bring GTC to eight cities around the world, Huang unveiled Xavier, our next-generation system-on-chip for powering self-driving cars; announced an agreement with TomTom, the Dutch mapping and navigation group, to use AI to create a cloud-to-car mapping system for self-driving cars; detailed our DriveWorks Alpha 1 release, and highlighted the work we're doing with some of Europe's most innovative startups and research labs. In the previous two weeks, Haung spoke at regional GTCs in Beijing and Taiwan that each drew crowds of more than 2,000.

Neural activity during cognitive tasks exhibits complex dynamics that flexibly encode task-relevant variables. Recurrent neural networks operating in the near-chaotic regime, which spontaneously generate rich dynamics, have been proposed as a model of cortical computation during cognitive tasks. However, existing methods for training these networks are either biologically implausible, and/or require a continuous, real-time error signal to guide the learning process. The lack of a biological learning method currently restricts the plausibility of recurrent networks as models of cortical computation. Here we show that a biologically plausible learning rule can train such recurrent networks, guided solely by delayed, phasic rewards at the end of each trial, for nontrivial tasks.