Honda and Google's parent company, Alphabet, are in formal talks to develop self-driving vehicles, the Japanese carmaker said on Thursday, months after the US firm signed a deal to use its technology in Fiat Chrysler minivans. The prospect of a deal between Honda and Alphabet's self-driving unit Waymo, which was spun off from Google earlier this month, is part of attempts by some car manufacturers to address the high cost of developing reliable automation software by teaming up with technology firms rather than going it alone. Honda, however, said any collaboration with Waymo did not mean it was abandoning efforts to develop its own autonomous driving system. While its driverless project has not garnered as much attention as similar plans by bigger firms such as Toyota, Honda unveiled a prototype driverless car in June and has said it hopes to see the fully autonomous vehicle appear on motorways in four years' time. "In addition to these on-going (in-house) efforts, this technical collaboration with Waymo could allow Honda research and development to explore a different technological approach to bring fully self-driving technology to market," Honda said in a statement.

There are still big questions to be answered about media agency practices, transparency and effectiveness. All the while, adtech technology and data management becomes more sophisticated. So, what does 2017 and beyond have in store? If we've learned anything over the last several years in programmatic it's that--in a world of commoditized inventory and 3rd party data--getting a programmatic edge requires diving deep into the data for insights. That means being better than your competitors at knowing where and how much to bid, which correlates directly with an organization's skill at data science.

Palantir CEO Alex Karp Says Going Public Is'A Possibility' Not to be a Grinch or a Scrooge, but you'd better watch out when it comes to some of this season's hottest Christmas toys. Last year, I wrote about Mattel's Hello Barbie, which responds realistically to your child by using natural language processing, machine learning and advanced analytics to parse what a child says and respond accordingly. In order to do this, the toy has to record what the child says, send it to a cloud-based server, and receive instructions back. On Barbie, there's a "listen" button the child has to press to start recording what she says in order to get a response. Mattel claims that only they and their tech partner (currently) have access to these recordings, but with the newest generation of talking toys, the audience is a bit larger.

The listing carries no penalties but will likely be an embarrassment for Alibaba, which has been trying to burnish its image in international markets. The move by the USTR comes even as the company claims to have used "big data" technologies to zero in, for example, on 13 factories and shops that were selling knockoff RAM modules under Kingston and Samsung brands, according to Alibaba's news hub Alizila. Its counterfeit goods monitoring and identification algorithm, for example, monitors about 100 dimensional characteristics, ranging from price to the online shops decorations, transaction records, product-release pattern and consumer complaints. Merchants and goods are rated on a 0 to 100 scale, with 80 usually treated as a red flag. The company also uses optical character recognition and the scanning and analysis of images and logos for suspicious listings.

A computer's ability to accurately identify images is a white whale for many technology companies, from Baidu to Google. One Australian startup has found a corner of the market to dominate, winning contracts with the European Union Intellectual Property Office (EUIPO) and IP Australia for algorithms that can detect and compare logos. SEE ALSO: Airbnb is getting into the airline booking disruption game with'Flights' TrademarkVision, which has support from Australia's CEA Startup Fund, uses machine learning to support image searches that can identify similar trademarks. Having a unique trademark or logo is vital, but many intellectual property registration bodies often require outdated forms of non-visual search that make comparison difficult. Australia, for example, relies on keywords, Europe on Vienna codes and the U.S. on design codes.

When we offered CS231n (Deep Learning class) at Stanford, we intentionally designed the programming assignments to include explicit calculations involved in backpropagation on the lowest level. The students had to implement the forward and the backward pass of each layer in raw numpy. This is seemingly a perfectly sensible appeal - if you're never going to write backward passes once the class is over, why practice writing them? Are we just torturing the students for our own amusement? Some easy answers could make arguments along the lines of "it's worth knowing what's under the hood as an intellectual curiosity", or perhaps "you might want to improve on the core algorithm later", but there is a much stronger and practical argument, which I wanted to devote a whole post to: In other words, it is easy to fall into the trap of abstracting away the learning process -- believing that you can simply stack arbitrary layers together and backprop will "magically make them work" on your data.

Financial Portfolio theories are one of the important achievements in financial economics in the last XX Century. One such theory goes by the designation of Markowitz Portfolio Theory or Modern Portfolio Theory, named after Nobel Prize in Economic Sciences winner Harry Markowitz. We read the Wikipedia entry for this theory and we can immediately confirm it as a mathematical and statistical theory at its core. And if it is mathematical and statistical at its core it is well positioned to be improved and enhanced by an algorithmic, computational approach. And that is the case with our paper's proposal: it is another one software approach to Portfolio Theory that turns the problem of finding the best efficient frontier predicted by the theory into a mathematical optimization problem, but from the new machine learning/deep learning perspective.

Real-time machine learning has access to a continuous flow of transactional data, but what it really needs in order to be effective is a continuous flow of labeled transactional data, and accurate labeling introduces latency. During this year, I heard and read a lot about real-time machine learning. People usually provide this appealing business scenario when discussing credit card fraud detection systems. They say that they can continuously update credit card fraud detection model in real-time (See "What is Apache Spark?", "…real-time use cases…" and "Real time machine learning"). It looks fantastic but not realistic to me.

This will reduce attrition, hiring costs and the effort.

With an increase in automation and the advent of artificial intelligence (AI), there are expected to be significant changes in the job market. The White House has released a report in the ways AI will affect the economy over the coming years and decades. The pace and direction of the development of AI will decide what sectors are affected and how soon. At a minimum, and in the near term future, drivers and cashiers are going to be replaced by machines. AI is going to sooner or later replace millions of jobs, and affect the livelihoods of these workers.