The success of deep learning in numerous application domains created the de- sire to run and train them on mobile devices. This however, conflicts with their computationally, memory and energy intense nature, leading to a growing interest in compression. Recent work by Han et al. (2015a) propose a pipeline that involves retraining, pruning and quantization of neural network weights, obtaining state-of-the-art compression rates. In this paper, we show that competitive compression rates can be achieved by using a version of soft weight-sharing (Nowlan & Hinton, 1992). Our method achieves both quantization and pruning in one simple (re-)training procedure. This point of view also exposes the relation between compression and the minimum description length (MDL) principle.

TL;DR A resilient Data Science Platform is a necessity to every centralized data science team within a large corporation. It helps them centralize, reuse, and productionize their models at peta scale. We've built Algorithmia Enterprise for that purpose. You've built that R/Python/Java model. "It started with your CEO hearing about machine learning and how data is the new oil. Someone in the data warehouse team just submitted their budget for an 1PB Teradata system, and the the CIO heard that FB is using commodity storage with Hadoop, and it's super cheap. A perfect storm is unleashed and now you have a mandate to build a data-first innovation team. You hire a group of data scientists, and everyone is excited and start coming to you for some of that digital magic to Googlify their business. Your data scientists don't have any infrastructure and spend all their time building dashboards for the execs, but the return on investment is negative and everyone blames you for not pouring enough unicorn blood over their P&L."

Video is the world's largest generator of data, captured by hundreds of millions of cameras in areas like government property, public transit, commercial buildings, and roadways. The number of cameras is expected to rise to more than one billion by 2020. AI technology enables powerful intelligent video analytics that can turn this staggering number of pixels into public safety and smart city solutions. NVIDIA Metropolis is the foundation of the AI City--an edge-to-cloud platform that can turn anonymized video into valuable insights. Driven by powerful technologies like NVIDIA Jetson TX2 at the edge and NVIDIA Tesla in the data center, it delivers intelligent video analytics for a wide range of applications.

Now that we have built a simple LSTM RNN network, how do we improve our error rate? Luckily for the open source community, many large companies have published the math that underlies their best performing speech recognition models. In September 2016, Microsoft released a paper in arXiv describing how they achieved a 6.9% error rate on the NIST 200 Switchboard data.

"Toronto and Canada for the past two decades has been at the forefront of AI, and that's the expertise we're bringing to Uber,"says Raquel Urtasun, who will lead Uber's Advanced Technologies Group in Toronto. Uber is launching a research group devoted to driverless car technology in Toronto, creating a third hub -- its first outside the U.S. -- for the company's ambitions in a frenzied field that Uber and its competitors believe will upend transportation, generating billions of dollars in the process. The Advanced Technologies Group will be led by Raquel Urtasun, a University of Toronto computer science professor who holds a Canada Research Chair in machine learning and computer vision. Urtasun uses artificial intelligence, particularly deep learning, to make vehicles and other machines perceive the world around them more accurately and efficiently. The group will hire "dozens" of researchers and engineers in the next few years, the company says.

It's a few years old, but still acts as a great technical introduction to the topic.

I keep changing my opinion on a daily basis, and I cannot seem to settle on one solid view of this puzzle. No, I am not talking about world politics or the current U.S. president, but rather something far more critical to humankind, and more specifically to our existence and work as engineers and researchers. I am talking about…deep learning. While you might find the above statement rather bombastic and overstated, deep learning indeed raises several critical questions we must address. In the following paragraphs, I hope to expose one key conflict related to the emergence of this field, which is relevant to researchers in the image processing community.

There is a tendency with any new technology to believe that it requires new management approaches, new organizational structures, and entirely new personnel. That impression is widespread with cognitive technologies -- which comprises a range of approaches in artificial intelligence (AI), machine learning, and deep learning. Some have argued for the creation of "chief cognitive officer" roles, and certainly many firms are rushing to hire experts with deep learning expertise. "New and different" is the ethos of the day. But we believe that successful firms can treat cognitive technologies as an opportunity to evolve or grow from previous work.

Just because Grindelwald and Dumbledore had a deadly brawl during their quest to revolutionize magic doesn't mean two great powers cannot be used in concert to change the world. This could be the worst way to start an important conversation about financial technology, but stick with me, it gets more interesting. We are speaking about the world-altering technology of Artificial Intelligence as the first superpower coupled with the financial system disruptive technology of cryptocurrency -- a decentralized payment system that circumvents government manipulation of currency and is forcing us to redefine the concept of money. The question is: Can these two technologies be used together to change the way ordinary people like you and me invest our money -- without expiring in a shower of blue sparks? But first, let's take a step back and look into them as individual concepts, with respect to their relationships to investment and trading.

After a quick overview of how Google has utilized its Artificial Intelligence (AI) technology, the article states, "Some companies have built their own AI research units and need to build highly customized models for specific applications. Yet, in doing so they quickly run up against the immense hardware requirements of building large deep learning models, often requiring entire accelerator farms for rapid iteration. In Google's case it offers a hosted deep learning platform called Cloud Machine Learning Engine that takes care of the hardware needs of deep learning development, allowing companies to focus on building their models and offload the computing requirements to Google. After all, few companies have invested so much in AI that they have built their own custom accelerator hardware like Google did with its Tensor Processing Units (TPUs)." Further into the article, the author, Kalev Leetaru, states analytics companies "are interested in building services for their customers, not conducting AI research. In following its externalization trend, Google has risen to this challenge by releasing many of its internal AI systems as public cloud APIs."