Without a doubt one of the most exciting potential uses for AI (Artificial Intelligence) and in particular deep learning is in healthcare. Traditionally, diagnosis of killer illnesses such as cancer and heart disease have relied on examinations of x-rays and scans to spot early warning signs of developing problems. Image recognition is of course one of the tasks at which deep learning excels – from Facebook's facial recognition to Google's image search, practical examples of it in use are becoming more common by the day. Although being able to tag pictures of our friends without typing their name, or find amusing images of cats when we want them, may seem trivial use cases, the same technology is quickly advancing to a point where more far-reaching implications are being realized. In China, lung cancer is the leading cause of death, claiming over 600,000 lives each year, largely due to high levels of air pollution.

IBM is trying to remove some of the complications related to image recognition with new tools to automate critical machine learning tasks. A major update of the company's PowerAI tools has a feature called AI Vision, an auto tuner that makes it easy to identify and classify pictures. It will also speed up image recognition by breaking down tasks over multiple clusters. AI Vision plays a big role in automating machine learning by creating a tuned model, said Sumit Gupta, vice president of machine learning. The software abstracts machine learning, and developers don't need knowledge of low-level access to frameworks to tune, train, and deploy image recognition models.

Can a machine learn aesthetics in a way a human would? Could it then look at a set of photos, and draw on those same aesthetics to reproduce a different set? It's a big question because it has long-term implications for how AI is going to develop. Is it just what you "like"? How does it all work? When you as a human find it hard to express what you like do you think a machine going to find it easy?

Google's Cloud Vision API is easily blinded by the addition of a little noise to the images it analyses, say a trio of researchers from the Network Security Lab at the University of Washington, Seattle. Authors Hossein Hosseini, Baicen Xiao and Radha Poovendran have hit arXiv with a pre-press paper titled Google's Cloud Vision API Is Not Robust To Noise (PDF) that says "In essence, we found that by adding noise, we can always force the API to output wrong labels or to fail to detect any face or text within the image." The authors explain that if one can add different types of noise to an image, the Cloud Vision API will always incorrectly analyse the pictures presented to it. The image at the top of this story (or here for m.reg readers) shows the false results the API returned. It doesn;t need to be a lot of noise: the authors found an average of 14.25 per cent "impulse noise" got the job done.

Editor's note: This post is part of our Trainspotting series, a deep dive into the visual and audio detection components of our Caltrain project. You can find the introduction to the series here. SVDS has previously used real-time, publicly available data to improve Caltrain arrival predictions. However, the station-arrival time data from Caltrain was not reliable enough to make accurate predictions. Using a Raspberry PiCamera and USB microphone, we were able to detect trains, their speed, and their direction. When we set up a new Raspberry Pi in our Mountain View office, we ran into a big problem: the Pi was not only detecting Caltrains (true positive), but also detecting Union Pacific freight trains and the VTA light rail (false positive).

Ensure that the product offerings on your pages have schema.org

Google is Trying to Turn Image Search into a Shopping Tool (Recode) Google has added a new shopping feature in Image Search called "style ideas" that shows users perusing fashion merchandise what specific items look like paired with others. Yext Shares Up Sharply In Initial Day of Trading, Portending Well for Local (Street Fight) Yext's shares jumped nearly 22% in the company's initial day of trading, with the price rising as high as $14.25 per share before settling to $13.41 at close. The strong opening was a hopeful message from Wall Street for the local marketing industry. Instagram's Snapchat Clone is Now More Popular than Snapchat (Quartz) In about eight months, Instagram Stories, the function of Facebook's image-sharing social network that allows users to post images and short videos that disappear after 24 hours, has amassed 200 million daily users, it announced today. Street Culture: Why Telecommuting Makes Sense for Many Tech Startups (Street Fight) "If you have the right team, the right employees, then they don't have to be there physically," says Kristen Stiles, co-founder and CEO of babysitter-finding app Sitter.me. "If you don't trust your employees to work at home, you shouldn't have hired them in the first place."

Google's latest AI toy may be its most clever: an automated drawing bot that analyzes what you're doodling in real time to suggest a more polished piece of clip art to replace it. Called AutoDraw, the software is another of Google's ongoing creative machine learning demonstrations that it releases as part of its AI Experiments series. It uses the underlying technology behind the company's experimental image recognition software to identify potential objects and pairs that with a database of neat and simplistic hand-drawn images. The company bills AutoDraw as a "drawing tool for the rest of us," and by us it means aesthetically impaired individuals who couldn't doodle themselves out of a paper bag. "AutoDraw pairs the magic of machine learning with drawings from talented artists to help you draw stuff fast," says the narrator in Google's AutoDraw teaser video.

This is another installment of Mighty AI's "Conversations in Machine Learning" blog series. Each week, our content human, Cassie, shares a summary of a recent conversation we had with a machine learning team and potential customer--what they're building, how they're handling training data today, etc. Read more about the series here. Each of those would be appropriate depending on the context--you'd connect its brand name to a search engine (because it is one, among other things), you'd call it a technology company with a glance-over of its corporate site, and you'd say it's a machine learning company if you dug into its products. That's because these days, machine learning is powering virtually everything the company does and is behind nearly everything it builds. To back up a bit, this is a European technology firm with deep roots in search and respectably deep roots in machine learning, too (employees there developed a proprietary ML methodology nearly a decade ago, and creating a whole new methodology sounds impressive to me at least).

Google just killed the Captcha, perhaps the most obstructive thing on the entire internet. For years, Captcha served as the primary way of telling humans and robots apart on the internet. It made sure that the person looking to access a website was actually a human being – ensuring that robots couldn't be used to send spam or flood a website with requests, for instance. But over time, robots have gradually become too clever for the often simple tests – which early on required people to transcribe hard-to-read text. With that, the technologies have become more complex, too.