As with the human brain, the neural networks that power artificial intelligence systems are not easy to understand. DeepMind, the Alphabet-owned AI firm famous for teaching an AI system to play Go, is attempting to work out how such systems make decisions. By knowing how AI works, it hopes to build smarter systems. But researchers acknowledged that the more complex the system, the harder it might be for humans to understand. The fact that the programmers who build AI systems do not entirely know why the algorithms that power it make the decisions they do, is one of the biggest issues with the technology.

Google's parent company Alphabet is increasingly commercializing the technology coming out of its artificial intelligence research unit DeepMind, four years after acquiring it. Earlier this week, Google's cloud business announced a new service that converts blocks of text into natural-sounding speech, the first product containing DeepMind code that's for sale. The new Google Cloud Text-to-Speech application programming interface costs $16 for every million characters of text it processes in DeepMind's artificial male and female voices. Alphabet operates other AI research groups, but DeepMind has been doing more futuristic work, like teaching computer systems to beat top-ranked players of the Chinese board game Go. DeepMind is one of Alphabet's so-called Other Bets, but in pushing the technology closer to Google, commercial applications are becoming more real.

The Linux Foundation is aiming to expedite the market for artificial intelligence (AI), machine learning and deep learning with the launch of the LF Deep Learning Foundation at the Open Networking Summit Monday. The LF Deep Learning Foundation is designed to be a neutral space for developers and data scientists in the open source community to work together on projects related to deep learning technologies. At launch, members of the group include Amdocs, AT&T, B.Yond, Baidu, Huawei, Nokia, Tech Mahindra, Tencent, Univa and ZTE. As part of the DL Foundation, the Linux Foundation is also launching the Acumos AI Project designed specifically for AI model discovery, development and sharing. AT&T Inc. (NYSE: T) and Tech Mahindra Ltd. are contributing the initial code to the project -- available today -- as part of a program they announced back in October.

One area that Deep Learning has not explored extensively is the uncertainty in estimates. Most Deep Learning frameworks currently focus on giving a best estimate as defined by a loss function. Occasionally something beyond a point estimate is required to make a decision. This is where a distribution would be useful. Bayesian statistics lends itself to this problem really well since a distribution over the dataset is inferred.

DeepMind's work is based on a solid grounding in neuroscience. This has underpinned the company's strategy of developing AI by teaching computers to mimic the thought processes of our own brains, in particular, how we use information to make decisions and learn from our mistakes. Google's interest in DeepMind likely lies in advancing the fields of machine learning and simulated neural networks – developing machines with more human-like thought processes, with the capacity to carry out jobs which previously would have required trained humans. Aside from winning at games, DeepMind has also been implemented across numerous healthcare projects, such as a collaboration with UCL's radiotherapy department to reduce the amount of time it takes to plan treatments. The group has said that by unleashing machine learning on the process of mapping a patient's head and neck area, the time taken to create treatment plans for these complicated procedures could be reduced from four hours to around one hour. One specific DeepMind project involves a collaboration with London's Moorfields Eye Hospital.

The question I always ask myself, just like any human being, is who am I and why do I exist? Who are we as humans and why do we exist? When I was in college, I had a much more naïve view. I was very much into computers and artificial intelligence, and I thought it must be the case that I'm destined to work on some computer algorithms and, along with my colleagues, figure out how the brain works and how the computer can be as smart as the brain, perhaps even become a substitute of the brain, and that's what artificial intelligence is about. That was the simplistic view that I had. I pursued that in my college, in my graduate years. I went to Carnegie Mellon and got a PhD in speech recognition, then went to Apple, then SGI, then Microsoft, and then to Google. In each of the companies, I continued to work on artificial intelligence, thinking that that was the pursuit of how intelligence worked, and that our elucidation of artificial intelligence would then come back and tell us, "Ah, that's how the brain works." We replicated it, so that's what intelligence is about. That must be the most important thing in our lives: our IQ, our ability to think, analyze, predict, understand--all that stuff should be explicable by replicating it in the computer. I've had the good fortune to have met Marvin Minsky, Allen Newell, Herb Simon, and my mentor, Raj Reddy. All of these people had a profound influence on the way I thought. It's consistent that they too were pursuing the understanding of intelligence.

As a beginner at deep learning, one of the things I realized is that there isn't much online documentation that covers all the deep learning tricks in one place. There are lots of small best practices, ranging from simple tricks like initializing weights, regularization to slightly complex techniques like cyclic learning rates that can make training and debugging neural nets easier and efficient. This inspired me to write this series of blogs where I will cover as many nuances as I can to make implementing deep learning simpler for you. While writing this blog, the assumption is that you have a basic idea of how neural networks are trained. An understanding of weights, biases, hidden layers, activations and activation functions will make the content clearer.

In 2014, the research paper Generative Adversarial Nets (GAN) by Goodfellow et al. was a breakthrough in the field of generative models. Leading researcher Yann Lecun himself called adversarial nets "the coolest idea in machine learning in the last twenty years." Today, thanks to this architecture, we're going to build an AI that generates realistic pictures of cats. To view the full working code, see my Github repository. It will help if you already have some experience in Python, Deep Learning and Tensorflow, and CNNs (Convolutional Neural Nets).

We're now in a phase of computing increasingly characterized by engineering complex systems of automation by applying several new and convergent technologies including: deep learning, cloud computing and massive arrays of data-producing sensors. NVIDIA's deep learning GPUs have been at the center of partnerships or agreements with a multitude of companies and customers. For example, NVIDIA partnered with U.S.-based Tesla Motors to provide deep learning GPUs for autonomous driving, Japan-based FANUC Robotics to create a deep learning "factory manager" to oversee manufacturing robots that manufacture other manufacturing robots, U.S.-based Amazon Web Services to provide a web platform to create deep learning applications, China-based Hikvision to power the company's sophisticated systems of video surveillance, and Japan-based Komatsu's SMARTCONSTRUCTION mining and construction equipment vehicles. If you stop to take account of how many deep learning partnerships NVIDIA has entered into with powerful commercial entities in various sectors of global industry, it quickly becomes overwhelming. But how does NVIDIA and its manufacturing partners apply deep learning along the supply chain that produces NVIDIA GPUs?

Artificial intelligence could make it a lot simpler to edit photos and videos--but could creativity get lost in the shuffle? Artificial intelligence is increasingly used to create art, such as images or music composed with machine learning. One of the prototypes, called Project Scene Stitch, illustrates how an algorithm could be used to replace ugly buildings in the foreground of a photo--a user would enter some key words, and the algorithm would find another image that would fit naturally into the space the user wanted to fill. Another prototype, Project Sky Replace, uses a deep-learning algorithm to remove the sky in a photo and replace it with other images of skies that match the one in the photo geometrically. It also considers the color balance of the foreground image and matches the foreground color to the sky color, making it possible to turn a photo of, for instance, the Eiffel Tower on a cloudy day into a properly lit sunset image. While the tools could be seen as antithetical to the creative process, Miller noted that they could add to overall creativity by freeing people from a lot of tedious work artists have to do today.