Everyone is buzzing about the impact of AI on work, and many leaders feel insecure about what it will mean in terms of their own career development and roles. Deep learning, machine learning, automation and robotics are creating a seismic shift across organizations. "We're now living in an age where [deep learning is] going to be mandatory for people building sophisticated software applications," according to Frank Chen, a partner at venture capital firm Andreessen Horowitz, who was quoted in a recent Fortune article. Soon, he notes, people will demand, "'Where's your natural-language processing version?' 'How do I talk to your app? Because I don't want to have to click through menus.'"

This folder contains two notebooks which demonstrate use of CTPN (Caffe implementation) [1,2] for Text box detection and CRNN (PyTorch implmentation) [3,4] for Text character recognition. Most online tutorials describe traditional OCR techniques using Tessaract. However Tessaract is not useful for scene text recognition, i.e. text occurring in natural scenes, with wide variation in fonts, colors and background. Over the last couple of years significant improvements have been made in using deep learning for OCR, in this demo we will show how you can use a textbox detector and a text recognition model to perform OCR on scene text. Its possible to get good out-of-box performance without any having to perform any fine-tuning.

Industries such as healthcare, government service, IT and telecommunication, media and advertising, BFSI, retail, travel, tourism, and hospitality create a huge amount of data base which is difficult to maintain by the conventional computing system. However with the introduction of artificial intelligence in these industries processing and managing of database became much efficient and rapid. Manufacturing is one of the first industry to take advantage of emerging AI technology, especially in the manufacturing process where robots were used to assemble and package products. Moreover, with the advent of technology, advanced robots will be able to perform complex operation in the manufacturing process such as assembling and testing of smart homes, smart city, vehicles, and electronics. Healthcare is another industry largely impacted by the deployment of AI technology. In fact, AI in healthcare industry would be the key area of contribution towards the'Fourth Industry Revolution'.

The work here presented is the result of a semester long independent research performed by Kenny Jones and Derrick Bonafilia (both Williams College 2017) under the guidance of Professor Andrea Danyluk. Kenny and Derrick are both heading to Facebook next year as Software Engineers and hope to continue studying GANs in whatever capacity is available to them. Generative Adversarial Networks (GANS) were introduced by Ian Goodfellow et. GANs address the lack of relative success of deep generative models compared to deep discriminative models. The authors cite the intractable nature of the maximum likelihood estimation that is necessary for most generative models as the reason for this discrepancy.

Escher may have just lost its lucrative stranglehold on the dorm room poster market thanks to artist Chris Rodley, who used a deep learning algorithm to merge a book of dinosaurs with a book of flower paintings. The results are magnificent, and deserve a spot on the walls of our finest art galleries. This isn't the first time Rodley has dabbled with a deep learning A.I. to create art. Using a website called Deepart.io, which is powered by an algorithm developed by Leon Gatys and a team from the University of Tübingen in Germany, Rodley previously merged a Trump family photo and various Muppet characters, with nightmarish results. The Deepart.io algorithm differs from what Google's Deep Dream does by applying features of an artist's visual style to another image, preserving recognizable details and features and using them to rebuild the target image from scratch.

Little wonder AI is sweeping the finance industry – it's hard to imagine an industry better suited for it. "Today the trade that I think is going to change the world, that every fund that every quant fund that wants to make money is adopting, is deep learning," Guarav Chakravorty, founder of qplum, an online advisory firm, said in conversation with our podcast's host, Michael Copeland for this week's episode of the AI Podcast While hedge funds have taken the lead in algorithmic investing – or robo-trading Chakravorty wants to bring that same machine learning investing approach to the rest of us. Qplum uses an AI robo-advisor that invests using algorithms to create customized portfolios. "We can use deep learning to make a level playing field where investing is a utility," he said. "Many people are not connected to investing because they don't know whether they're going to do it right, they don't know whether they know enough."

A generative adversarial network (GAN) is composed of two separate networks - the generator and the discriminator. It poses the unsupervised learning problem as a game between the two. In this post we will see why GANs have so much potential, and frame GANs as a boxing match between two opponents. Deep learning is famously biologically inspired and many of the major concepts in deep learning are intuitive and grounded in reality. The fundamental truth of deep learning is that it's hierarchical -- the layers in a network and the representations they learn build on each other.

Deep learning is a subfield of machine learning and it comprises several approaches to tackling the single most important goal of AI research: allowing computers to model our world well enough to exhibit something like what we humans call intelligence. On a basic conceptual level, deep learning approaches share a very basic trait. DL algorithms interpret the raw data through multiple processing layers. Each of these layers takes the output of the previous one as its input and creates a more abstract representation of it. As a result, the more data is being fed into the right algorithm, the more general are the rules and features that it's able to infer in relation to a given scenario and, therefore, the apter it gets at handling new, similar situations.

Recently, Lab41 teamed up with Cyber Reboot (a sister lab) to explore the intersection of deep learning (DL) and cyber security in a software defined network (SDN) environment. We called it Poseidon, based heavily on it being a cool word with the letters s, d, and n in order. The goal was to use predictions about network traffic to automatically update a network's posture. This entailed three main objectives: performing deep learning on packet data, setting up an SDN environment, and scheduling a microservice to connect the two (for more information and code visit our Github page). Since I belong to the cult of deep learning, I was tasked with the first objective.