Deep learning with neural networks is applied by an increasing number of people outside of classic research environments, due to the vast success of the methodology on a wide range of machine perception tasks. While this interest is fueled by beautiful success stories, practical work in deep learning on novel tasks without existing baselines remains challenging. This paper explores the specific challenges arising in the realm of real world tasks, based on case studies from research \& development in conjunction with industry, and extracts lessons learned from them. It thus fills a gap between the publication of latest algorithmic and methodical developments, and the usually omitted nitty-gritty of how to make them work. Specifically, we give insight into deep learning projects on face matching, print media monitoring, industrial quality control, music scanning, strategy game playing, and automated machine learning, thereby providing best practices for deep learning in practice.

We introduce a new entity typing task: given a sentence with an entity mention, the goal is to predict a set of free-form phrases (e.g. skyscraper, songwriter, or criminal) that describe appropriate types for the target entity. This formulation allows us to use a new type of distant supervision at large scale: head words, which indicate the type of the noun phrases they appear in. We show that these ultra-fine types can be crowd-sourced, and introduce new evaluation sets that are much more diverse and fine-grained than existing benchmarks. We present a model that can predict open types, and is trained using a multitask objective that pools our new head-word supervision with prior supervision from entity linking. Experimental results demonstrate that our model is effective in predicting entity types at varying granularity; it achieves state of the art performance on an existing fine-grained entity typing benchmark, and sets baselines for our newly-introduced datasets. Our data and model can be downloaded from: http://nlp.cs.washington.edu/entity_type

To put it mildly, news media has been on the sidelines in AI development. As a consequence, in the age of AI-powered personalized interfaces, the news organizations don't anymore get to define what's real news, or, even more importantly, what's truthful or trustworthy.

This Saturday, the Robot Film Festival is taking place in Portland, Ore. This is the 8th year of the festival, and after bouncing around between San Francisco, Pittsburgh, and Los Angeles, the festival has (at least temporarily) settled on the greatest city on Earth (and coincidentally my hometown), Portland. The theme this year is "Vintage Arcade Revival," celebrating the culture and technology of the 1980s. Or at least, celebrating what passed for culture and technology back then. I'll be giving a short introductory talk to help kick off the film festival, about the progress we've seen in social home robots from the 1980s to today.

Jefferson Graham shows the many options for finding audio podcasts to listen to, from apps to connected speakers, TV and the car, on #TalkingTech. In July, we've covered everything from how to get the beat deals on Amazon Prime Day to tips on creating and publishing an e-book and we previewed the new website with Netflix's Phil Rosenthall on Talking Tech, USA TODAY's daily, seven-days-a-week podcast. You can listen to Talking Tech on Stitcher, Apple or Google Podcasts or wherever you enjoy online audio. Talking Tech offers the latest tech news updates, gadget reviews, opinion on tech trends and interviews with insiders. Here are the July Talking Tech episodes.

Thursday, Thomson Reuters launched Westlaw Edge, an updated, artificial intelligence-assisted legal research platform. The updates include new warnings for invalid or questionable law, litigation analytics, a tool to analyze statutory changes and an improved AI-enhanced search called WestSearch Plus. Wednesday, Thomson Reuters hosted a group of journalists at its Times Square office to demo and illustrate the new version of Westlaw. "We're swinging for the fences, we're going after harder problems," said Mike Dahn, senior vice president of Westlaw product management. Company representatives would not disclose the cost of the new offering.

Netflix standardizing on a proven cloud platform and collaborating with Amazon Web Services development teams on machine-learning security initiatives and many others is proving to be a powerful catalyst fueling the subscription services' global growth. Netflix's exponential growth this year is attributable in part to the cloud platform decisions made years ago that enable their subscription-based business model to scale globally securely. Last year at Amazon's AWS re:Invent 2017 Conference, Greg Peters, Chief Product Officer of Netflix provided insights into how closely Netflix and AWS work together to create innovative new services based on AWS' advances in machine learning-powered security, developer apps, and scalability. It's an insightful session into how Netflix is relying on Amazon to do the heavy lifting of infrastructure development and can be viewed here, AWS re:Invent 2017 - Fireside Chat: Steve Schmidt, Jenny Brinkley, and Greg Peters of Netflix. AWS and Netflix development teams are using machine learning-powered security to analyze data access patterns and look for anomalous account activity.

Hewes Pictures has announced the theatrical and digital release of the anthology film A.I Tales. Check out a few trailers for it and some images below. A.I Tales is a collection of short sci-fi stories from short film distributor Hewes Pictures is being released theatrically by Freedom Cinema, a jumpstart anthology film label. The film is made up of futuristic, high concept stories about artificial intelligence. Whether it's love found in time of over-population or exploration of the unknown and space-travel, "A.I. Tales has a bit of something for every fan of the genre. Altogether, these stories provide a one-of-a-kind experience and a unique view of the near future."

This post was done in partnership with Wirecutter. When readers choose to buy Wirecutter's independently chosen editorial picks, it may earn affiliate commissions that support its work. An upgraded version (with Alexa) of our top pick, the N79, in our robot vacuum guide, the N79S is usually well over $200. If you use Alexa voice commands for home automation, it's a nice option. Use code SQOG2YJD in cart to drop the current sale price of $180 to $170 (substantially cheaper than the current price of its sibling, the N79), which matches the low we saw for it Monday when it was a deal of the day.

Season 5 of Fortnite has officially begun, and with it comes a brand new Battle Pass. The Battle Pass is Fortnite: Battle Royale's most ingenious innovation. For 950 V-Bucks, or just about $10 USD, you buy into the paid version of the game for one season. Seasons last 10 weeks and come with weekly challenges, 100 cosmetic rewards and lots more. So now that the Season 5 Battle Pass has gone live, let's take a look at what's in it.