At Bossa Nova we create service robots for the global retail industry. Our robots' mission is to make stores run efficiently by automating the collection and analysis of on-shelves inventory data in large scale stores. Navigating smoothly along the aisles, we circulate autonomously among busy customers and employees. If we were a self- driving car we'd be operating at level 5 autonomy. Yep, it is possible to move, scan and analyze all at the same time.

AI is the compelling topic of tech conversations du jour, yet within these conversations confusion often reigns – confusion caused by loose use of AI terminology. The problem is that AI comes in a variety of forms, each one with its own distinct range of capabilities and techniques, and at its own stage of development. Some forms of AI that we frequently hear about, such as Artificial General Intelligence, the kind of AI that might someday automate all work and that we might lose control of – may never come to pass. Others are doing useful work and are driving growth in the high performance sector of the technology industry. These definitions aren't meant to be the final word on AI terminology, the industry is growing and changing so fast that terms will change and new ones will be added.

Whether you're new to deep learning or want to build advanced deep learning projects in the cloud, it's easy to get started by using AWS. The AWS Deep Learning AMIs, available in both Ubuntu and Amazon Linux versions, let you run deep learning applications in the cloud at any scale. The Amazon Machine Images (AMIs) come with pre-installed, open source deep learning frameworks including Apache MXNet, TensorFlow, the Microsoft Cognitive Toolkit (CNTK), Caffe, Caffe2, Theano, Torch, and Keras. With the AMIs, you can train custom AI models, experiment with new algorithms, and learn new deep learning skills and techniques. There is no additional charge to use the AMIs--you pay only for the AWS resources needed to store and run your applications.

Deep learning would be part of every developer's toolbox in near future. It wouldn't just be tool for experts. In this course, we will develop our own deep learning framework in Python from zero to one whereas the mathematical backgrounds of neural networks and deep learning are mentioned concretely. Hands on programming approach would make concepts more understandable. So, you would not need to consume any high level deep learning framework anymore. Even though, python is used in the course, you can easily adapt the theory into any other programming language.

A long-standing problem at the interface of artificial intelligence and applied mathematics is to devise an algorithm capable of achieving human level or even superhuman proficiency in transforming observed data into predictive mathematical models of the physical world. In the current era of abundance of data and advanced machine learning capabilities, the natural question arises: How can we automatically uncover the underlying laws of physics from high-dimensional data generated from experiments? In this work, we put forth a deep learning approach for discovering nonlinear partial differential equations from scattered and potentially noisy observations in space and time. Specifically, we approximate the unknown solution as well as the nonlinear dynamics by two deep neural networks. The first network acts as a prior on the unknown solution and essentially enables us to avoid numerical differentiations which are inherently ill-conditioned and unstable. The second network represents the nonlinear dynamics and helps us distill the mechanisms that govern the evolution of a given spatiotemporal data-set. We test the effectiveness of our approach for several benchmark problems spanning a number of scientific domains and demonstrate how the proposed framework can help us accurately learn the underlying dynamics and forecast future states of the system. In particular, we study the Burgers', Korteweg-de Vries (KdV), Kuramoto-Sivashinsky, nonlinear Schr\"{o}dinger, and Navier-Stokes equations.

"Disguised face identification" technology could make staying anonymous in public a bigger challenge (and make it easier to digitally unmask dissenters). To evade the risk of being IDed by authorities, protesters, looters, and rioters may obscure their faces with hats, scarves, or sunglasses. But for better or worse, dissenting in disguise may no longer mean dissenting anonymously. New research introduces an artificial intelligence-powered framework for "disguised face identification" – abbreviated by the researchers as DIC. The work, which considered how to identify masked faces in a crowd, addresses two related concerns at once: Most facial recognition technologies not only lack disguised-detection abilities, they also struggle to separate individuals from their backgrounds.

Ever since computer scientist Alan Turing first proposed his famous test of machine intelligence in 1950, the question of what it means for a computer to think has revolved around one basic question: Can it imitate a human's own thinking so closely that nobody can tell the difference? At first glance, the imitation going on with the latest A.I. creation from Microsoft's Deep Learning Technology Center is of cameras, not the human mind. The bot works to create photorealistic images -- in this case, mostly of birds -- using nothing but text descriptions and a huge repository of similar photographs to draw on. The bird pictured up top is real. The one below is not.

With new neural network architectures popping up every now and then, it's hard to keep track of them all. Knowing all the abbreviations being thrown around (DCIGN, BiLSTM, DCGAN, anyone?) can be a bit overwhelming at first. So I decided to compose a cheat sheet containing many of those architectures. Most of these are neural networks, some are completely different beasts. Though all of these architectures are presented as novel and unique, when I drew the node structures… their underlying relations started to make more sense.

What happens when you take a perfectly good neural network and, figuratively, stick a screwdriver in its brain? You get melancholy glitch-art music videos that turn talking heads into digital puppets. A machine learning developer named Jeff Zito made a series of music videos using a deep learning network based on Face2Face. Originally developed to generate stunningly realistic image transfers, like controlling a digital Obama in real-time using your own facial movements, this project takes it in a different direction. When it comes to art, for example, computations and algorithms often don't matter as much as chaos and noise do.

The Data Science Bowl is underway again, and this year, deep learning is the game. For the next 90 days, data scientists will have the chance to submit algorithms that can identify nuclei in cell samples without human intervention. The idea is to speed up drug target identification by tasking a deep learning model with analyzing millions of cell samples, rather than relying on human scientists. "All current options for nuclei detection require time-consuming biologist intervention. There are no deep learning models available today that can identify nuclei across multiple experimental setups and testing conditions," the event's official statement says.