The number of companies using machine learning is accelerating, but so far there are no tools to validate, verify and debug these systems. That presents a problem for the chipmakers and systems companies that increasingly rely on machine learning to optimize their technology because, at least for now, it creates the potential for errors that are extremely difficult to trace and fix. At the same time, it opens up new opportunities for companies that have been developing static tools to expand their reach well beyond just the chip, where profits are being squeezed by system vendors. But as shown in part one of this series, that will take years rather than months to fix. Research is just beginning on how to tackle these problems, let alone develop comprehensive tool suites. "Across the board, machine learning is suddenly becoming very interesting," said Sundari Mitra, CEO and co-founder of NetSpeed Systems.

Salesforce officially unveiled its artificial intelligence (AI) capability called Einstein yesterday -- just hours before Oracle kicked off OpenWorld. Constellation Research analyst Doug Henschen said it was no coincidence that Salesforce made a point of introducing its AI capabilities just ahead of Oracle's big user conference. "And likewise, it's no coincidence that we'll hear AI announcements from Oracle this year. Its work has been stealthy while Salesforce has very publicaly been loading up on machine learning and AI-related acquisitions," he said. Salesforce promises Einstein will forever change customer relationship management (CRM).

The near future of artificial intelligence (AI) won't be defined by ushering in a race of sentient machines. While we're inching closer to the goal of AI brain mapping, the next era of AI will be more about imbuing the software and applications we use every day with deep learning, machine learning, predictive analytics, and natural language processing (NLP). Those capabilities will run under the surface, along with serving as tools upon which to build. It's about making AI a given rather than a novelty. Customer relationship management (CRM) giant Salesforce unveiled its plan for more accessible, natively integrated AI for businesses today with the announcement of Salesforce Einstein, its "AI for CRM" technology.

Usually, host Rose Eveleth spends Flash Forward imagining a future made possible by new paradigms, like universal body-hacking or telepathy. This week, though, she gave the creative responsibilities to a recurrent neural network, who offered up a script based on every previous episode of her podcast, plus The War of the Worlds and The Hitchhiker's Guide to the Galaxy radio play. As a result, Eveleth speculates about two futures: a post-apocalyptic world where witches go to space, and our possible future where AI regularly create our entertainment.

A Maine high school unable to fill a vacant teacher position has turned to a foreign language computer program to educate students. With money already earmarked for the job, The Morning Sentinel reports Madison Area Memorial High School opted to purchase the Rosetta Stone program to serve as its full-time French and Spanish teacher. Principal Jessica Ward says the situation isn't perfect, but Rosetta Stone was the best option moving forward this year. The school was forced to purchase the program, which is currently used in more than 4,000 schools nationwide, when no one applied. Rosetta Stone officials say the program is ideally used in conjunction with a live teacher.

No doubt, this is the best-known line from Mario Puzo's book, The Godfather (1969) and the Oscar-award winning film of the same title (1972). Perhaps it may be one of the best-known lines in any film and ranks second only to "Frankly, my dear, I don't give a damn" as the most celebrated quotation from an American film. Of course, the'offer' here is "do as I say or I'll kill you"… Corleone's preferred method of intimidating those around him and assuring that he gets exactly what he wants in an expeditious fashion. So you may be asking, what does this famous gangster line have to do with the fast emerging field of "bots" – software applications that run automated tasks and scripts over the Internet. Quite frankly, some people have a healthy fear of Artificial Intelligence (AI) and the potential for unintended consequences of a future where machines have taken control of the world and reign over humans (think about the plot of The Matrix or The Terminator). So the purpose of this blog is to take a deeper look at the current state of AI applications on the Internet and to some extent, demystify what is going on with bots.

In a previous tutorial series I went over some of the theory behind Recurrent Neural Networks (RNNs) and the implementation of a simple RNN from scratch. That's a useful exercise, but in practice we use libraries like Tensorflow with high-level primitives for dealing with RNNs. With that using an RNN should be as easy as calling a function, right? In this post I want to go over some of the best practices for working with RNNs in Tensorflow, especially the functionality that isn't well documented on the official site. Check out the tf.SequenceExample Jupyter Notebook here! RNNs are used for sequential data that has inputs and/or outputs at multiple time steps.

Nobody understands why deep neural networks are so good at solving complex problems. Now physicists say the secret is buried in the laws of physics. In the last couple of years, deep learning techniques have taken the world of artificial intelligence by storm. One by one, the abilities and techniques that humans once imagined were uniquely our own have begun to fall to the onslaught of ever more powerful machines. Deep neural networks are now better than humans at tasks such as face recognition and object recognition.

Webroot has announced the acquisition of CyberFlow Analytics, a firm which specializes in harnessing machine learning technology to automatically detect cybersecurity threats. On Monday, the Broomfield, CO-based company said the move enhances Webroot's "ability to address the explosion of internet-connected devices and an increasingly complex threat landscape." Financial details of the acquisition were not disclosed. CyberFlow Analytics' FlowScape technology is a highlight of the acquisition. The SaaS-based cybersecurity solution, available in versions suitable for the enterprise and SMBs, implements machine learning to sift through network noise to detect patterns and network anomalies associated with potential threats.

It's what we feel are the most important reads.