This is the second of a multi-part series explaining the fundamentals of deep learning by long-time tech journalist Michael Copeland. That's how to think about deep neural networks going through the "training" phase. Neural networks get an education for the same reason most people do -- to learn to do a job. More specifically, the trained neural network is put to work out in the digital world using what it has learned -- to recognize images, spoken words, a blood disease, or suggest the shoes someone is likely to buy next, you name it -- in the streamlined form of an application. This speedier and more efficient version of a neural network infers things about new data it's presented with based on its training.

American space agency NASA has built an Artificial Intelligence (AI) platform capable of aiding firefighters when they enter a burning building. The platform, called Audrey, is the product of a partnership between the agency's Jet Propulsion Lab (JPL) and the Department for Homeland Security (DHS). This project forms part of the Next Generation First Responder program, which aims to identify ways firefighters, police and paramedics can stay safe while in the field. Audrey collects data about heat, gases and other signs of danger to help first responders get through the flames safely and quickly, letting them save victims. To make the AI platform possible, the designers used several technologies developed by NASA and the Department of Defence.

Microsoft on Monday announced it bought a startup to boost its artificial intelligence capabilities, and rival Apple confirmed it has boosted... Microsoft did not disclose financial terms of its deal to buy Genee, which specialises in using machine smarts to handle the time-sucking task of scheduling meetings. "Genee uses natural language processing and optimised decision-making algorithms so that interacting with a virtual assistant is just like interacting with a human one," Outlook and Office 365 corporate vice president Rajesh Jha said in a blog post. For example, Genee can be copied into an email exchange to act as a virtual assistant of sorts to pin down a time for a business or social meeting. Jha touted Genee as having designed "an intelligent virtual assistant specialised in the appointment decision." The technology was expected to be put to work in Office 365 software that Microsoft offers as a service in the internet cloud.

What are you doing yourself? We work for the Chief Data Officer of a CAC40 company who even decided to make his 2017 tagline as "Zero POC!". Machine Learning and Artificial Intelligence are not anymore reserved to Amazon, Facebook or Google. However, very few initiatives (less than 1/3rd) deliver some measurable ROI. To overcome this hurdle, it is however somewhat rather simple. For a change, do not start only investing hundred thousands or millions euros building a data lake… We quite often see that a couple of hundreds megabytes already offer very rich insights to train predictive models. And plugging a simple but well designed bot with powerful NLP (Natural Language Processing) to your legacy BI datawarehouses free up tremendous hidden value.

For most consumers, push notifications are more an annoyance than an effective marketing or sales tool, with the typical push notification only converting about 4–8 percent of recipients. That means that, for every 100 push notifications a brand sends out, only about four to eight of them wind up being at all successful at enticing or engaging potential consumers. Consumers simply don't engage with the average push notification that pops up on their phone. But Vizury, an omnichannel marketing platform based out of Bangalore, India, is seeking to change all that by using machine learning and Big Data analytics to help make push notifications more effective, engaging and viewed, instead of just ignored by consumers. Vizury recently introduced its latest platform, Engage Commerce, designed to spike user retention and conversion rates by blending machine learning and analytics to "deliver deeply personalized one-to-one recommendations" that can be sent individually to target specific and unique users.

The battle over how to make our lives more productive continues, especially on the email front. Baydin has spent the past six years helping people bring a degree of normalcy to their inbox through its Boomerang for Gmail plugin. This takes a scheduling approach -- you can specify when you want to send or receive email. But solving distribution is just one part of the equation, as the more important problem is how to create impactful communication that requires fewer emails. Enter Boomerang Respondable, a plugin for Gmail and Outlook that uses artificial intelligence (A.I.) to predict email replies and then makes suggestions about how to improve your writing and strike the right tone.

While many large technology and car manufacturers are building self-driving cars from the ground up, an increasing number of off-the-shelf systems will allow plenty of other models to take to the road without a driver. The latest such project is a newly announced partnership between the GM tech spinoff Delphi Automotive and Israeli machine vision company Mobileye. Both companies are no stranger to self-driving car technology: they both supply sensors and software to big-name automakers, including the technology behind Volvo's vehicle detection systems and, until recently, Tesla's Autopilot. But while both companies work closely with car manufacturers--Mobileye is working with BMW to put an autonomous car on the road by 2021, for instance--other large automakers, such as Ford, are building systems in-house. Now, according to the Wall Street Journal, the pair plans to invest "several hundred million dollars" in developing an off-the-shelf autonomous driving system, presumably for use by automakers who don't have the capacity or inclination for such research and development. The collaboration promises to demonstrate a system that will allow cars to autonomously navigate challenging road conditions--such as roundabouts or turns across multiple traffic lanes--as soon as January.

For IT organizations, machine learning is looking like an essential capability in the decade ahead. For the past few months, Google CEO Sundar Pichai has been extolling the value of AI and machine learning to his company. Gartner has added machine learning to its 2016 Hype Cycle, putting it at the peak of inflated expectations. The Hype Cycle, said Gartner research director Mike J. Walker in a statement, lists technologies that show "promise in delivering a high degree of competitive advantage over the next five to 10 years." Now, researchers at the Stanford University School of Medicine have demonstrated trained computers can outperform doctors when evaluating the slides of lung cancer patients, a finding which underscores the value of machine learning for data analysis tasks involving image recognition.

VCs are getting serious about space-related startups, and there are some truly exponential growth opportunities in the SpaceTech ecosystem. But the majority of money that went into SpaceTech last year was in just two deals -- a 1 billion fundraise for SpaceX and a 500 million raise for OneWeb. If you ignore those as outliers, SpaceTech funding in 2015 was only around 300 million. The segment of startups seeing the most VC attention is the earth observation (EO) segment. Companies building earth observation (EO) satellite constellations (basically, cameras put into orbit and photographing the Earth on a regular basis) pulled in more than half the SpaceTech funding from 2012 to 2015 and 72% in 2015.

Somewhere at Google, researchers are blurring the line between reality and fiction. Tell me if you've heard this one, Silicon Valley fans -- a small team builds a neural network for the sole purpose of making media files teeny-tiny. Google's latest experiment isn't exactly the HBO hit's Pied Piper come to life, but it's a step in that direction: using trained computer intelligence to make images smaller than current JPEG compression allows. Google's approach relies on forcing its network to learn compression the hard way. Researchers sampled six million compressed photos from the internet and broke them each 32 x 32 pixel pieces.