In late December 2016, I announced a small gift of 10 Deep Learning books to people interested in or working in AI. This is my way of paying back to the community which has been extremely generous with ideas and code. I asked people to send me an email letting me know their interest in AI and their contributions to the community. Here is the video of the announcement. I received nearly 300 emails from people from around the world -- pretty much every continent other than Antarctica!

Artificial intelligence had a big year in 2016, with massive strides forward in deep learning and intuitive interfaces. On the consumer level, digital assistants like Siri and Cortana significantly improved, and in higher circles, AlphaGo definitively beat a human master Go player for the first time in history. As with most technology, AI capabilities have been evolving at an astounding rate, far surpassing expectations and setting us up for a landmark year in 2017. So how exactly will AI grow in 2017, and how should business owners and marketers prepare accordingly? First off, there's the Internet-of-Things (IoT), which has been on the verge of taking off for a few years now.

"At Altek, we are constantly striving to enhance our digital image solutions and set the direction for the future of smarter imaging devices," said Jason Lin, General Manager and Corporate Senior Vice President of Altek. "CEVA's imaging and vision DSP provides the platform which allows us to further enhance the image quality of our solutions and push the boundaries of what a camera can do using artificial intelligence and advanced vision algorithms." "Altek is a proven leader in imaging, with a strong track record in the smartphone space and we are excited to work with them," said, Ilan Yona, vice president and general manager of CEVA's Vision Business Unit. "The combination of Altek's advanced imaging technologies along with our DSP-based vision and machine learning offering creates one of the most intelligent digital imaging solutions on the market today." CEVA's latest generation imaging and vision DSP platforms address the extreme processing requirements and low power constraints of the most sophisticated machine learning and machine vision applications used in smartphones, surveillance, augmented reality, sense and avoid drones and self-driving cars.

Intelligent machines powered by artificial intelligence (AI) computers that can learn, reason and interact with people and the surrounding world are no longer science fiction. Thanks to a new computing model called deep learning using powerful graphics processing units (GPUs), AI is transforming industries from consumer cloud services to healthcare to factories and cities. Many of these are in place already, providing new services to millions around the world. However, no industry is poised for such a significant change as the $10 trillion transportation industry. The automotive market is next, and the opportunity to develop advanced self-driving vehicle holds the promise to the world of dramatically safer driving and new mobility services.

Here is our selection of featured articles and resources posted since Monday. This is our first "Thursday News" of the year, and the quality of the articles submitted and accepted was especially high.

Like the course I just released on Hidden Markov Models, Recurrent Neural Networks are all about learning sequences - but whereas Markov Models are limited by the Markov assumption, Recurrent Neural Networks are not - and as a result, they are more expressive, and more powerful than anything we've seen on tasks that we haven't made progress on in decades. So what's going to be in this course and how will it build on the previous neural network courses and Hidden Markov Models? In the first section of the course we are going to add the concept of time to our neural networks. I'll introduce you to the Simple Recurrent Unit, also known as the Elman unit. We are going to revisit the XOR problem, but we're going to extend it so that it becomes the parity problem - you'll see that regular feedforward neural networks will have trouble solving this problem but recurrent networks will work because the key is to treat the input as a sequence.

Deep Learning is a hot buzzword of today. The recent results and applications are incredibly promising, spanning areas such as speech recognition, language understanding and computer vision. Indeed, Deep Learning is now changing the very customer experience around many of Microsoft's products, including HoloLens, Skype, Cortana, Office 365, Bing and more. Deep Learning is also a core part of Microsoft's development platform offerings with an extensive toolset that includes: the Microsoft Cognitive Toolkit, the Cortana Intelligence Suite, Microsoft Cognitive Services APIs, Azure Machine Learning, the Bot Framework, and the Azure Bot Service. Our Deep Learning based language translation in Skype was recently named one of the 7 greatest software innovations of the year by Popular Science, and this technology has now helped machines achieve human-level parity in conversational speech recognition.

There are certain practices in Deep Learning that are highly recommended, in order to efficiently train Deep Neural Networks. In this post, I will be covering a few of these most commonly used practices, ranging from importance of quality training data, choice of hyperparameters to more general tips for faster prototyping of DNNs. Most of these practices, are validated by the research in academia and industry and are presented with mathematical and experimental proofs in research papers like Efficient BackProp(Yann LeCun et al.) and Practical Recommendations for Deep Architectures(Yoshua Bengio). A lot of ML practitioners are habitual of throwing raw training data in any Deep Neural Net(DNN). And why not, any DNN would(presumably) still give good results, right?

Deep Learning has been the core topic in the Machine Learning community the last couple of years and 2016 was not the exception. In this article, we will go through the advancements we think have contributed the most (or have the potential) to move the field forward and how organizations and the community are making sure that these powerful technologies are going to be used in a way that is beneficial for all. One of the main challenges researchers have historically struggled with has been unsupervised learning. We think 2016 has been a great year for this area, mainly because of the vast amount of work on Generative Models. Moreover, the ability to naturally communicate with machines has been also one of the dream goals and several approaches have been presented by giants like Google and Facebook.

If artificial intelligence has its way, discounting could disappear, thanks to software that tells retailers exactly what and how many products to buy, and when to put them on sale to sell them at full price. Online shopping could become a conversation, where the shopper describes the dress of their dreams, and, in seconds, an AI-powered search engine tracks down the closest match. Designers, merchandisers and buyers could all work alongside AI, to predict what customers want to wear, before they even know themselves. In the last few years, a trifecta of cheap, ubiquitous, powerful computing; big data; and the development of deep learning have triggered a revolution in artificial intelligence. The computing devices that now fill our everyday lives generate large data sets, which "deep learning" algorithms analyse to find trends, make predictions and perform specific tasks, such as identifying specific objects in an image.