The most important thing for executives is to just start engaging with AI today – tomorrow is not good enough," warns Shamus Rae, partner and head of digital disruption at KPMG UK. "Business leaders need to understand the capabilities of these new technologies and put in place an AI strategy that includes some clear self-challenge. This ensures that they don't get stuck in'play' mode and fail to make any tangible change to their operations or business model." Mr Rae predicts that, due to the common misconception of AI, "we will see some major missteps by household names failing to adapt fast enough". "Think hard about what problem in your business you want to solve, not with artificial intelligence but with data," advises Kim Nilsson, founder and chief executive of data science hub Pivigo. "The solution always needs to come from that intersection of where your business challenges overlap with available data sets.

Are we designing AI or is AI designing us? What does AI have anything to do with Art which is innately a human quality? Harshit takes us to a journey of his experiments with Art and AI and the results are fascinating. Harshit Agrawal is fondly known as first AI artist from India. He is also HCI researcher, poet and traveler who builds tools to study how technology can help enhance human creative expression.

Deepfakes have made their way into the radar of much of the First World. As with many technology phenomena, deepfakes have their origins in pornography – editing (the Reddit page that originally popularized deepfakes was banned in early 2018). In April of this year, I was asked by UNICRI (the crime and justice wing of the UN) to present the risks and opportunities of deepfakes and programmatically generated content at United Nations headquarters for a convening titled: Artificial Intelligence and Robotics: Reshaping the Future of Crime, Terrorism, and Security. Instead of speaking about the topic, we decided it would be better to showcase the technology to the UN, IGO, and law enforcement leaders attending the event. So we took a video of UNICRI Director Ms. Bettina Tucci Bartsiotas, and created a deepfake, altering her words and statements by using a model of her face on another person.

A new survey of software engineers from careers firm Hired shows they want to learn machine learning, love Python, hate PHP and make a lot of money, especially in San Francisco where search engineer salaries average about $157,000. Furthermore, according to the 2019 State of Software Engineers Report, blockchain engineers are highly sought after, seeing a huge 517 percent year-over-year increase in demand, far exceeding the 132 percent increase in demand for No. 2, security engineers. Hired said it publishes the report to fuel career conversations among developers and to provide data they can use to achieve goals. Topping the list of those goals is learning machine learning. "Machine learning is the No. 1 technology engineers want to learn," the report said.

A machine learning algorithm that can predict the mechanical properties of metal–organic frameworks (MOFs) offers a way to overcome these highly varied and versatile materials' achilles heel – their instability.1 The team behind this work hope that this computational tool will speed up acceptance of these materials by industry. MOFs are a type of crystalline coordination polymers that form porous structures by combining metal clusters and organic ligands. 'Their "building block" nature allows chemists to easily tune their syntheses to tailor the pore size and surface chemistry for a specific application,' explains David Fairén-Jiménez at the University of Cambridge, UK. 'However, if you wish to use MOFs in real life, you need to shape them into pellets, and this densification may destroy their porosity, thus their functionality.'

On Tuesday, in an 8-1 tally, the San Francisco Board of Supervisors voted to ban the use of facial recognition software by city departments, including police. Supporters of the ban cited racial inequality in audits of facial recognition software from companies like Amazon and Microsoft, as well as dystopian surveillance happening now in China. At the core of arguments around the regulation of facial recognition software use is the question of whether a temporary moratorium should be put in place until police and governments adopt policies and standards or it should be permanently banned. Some believe facial recognition software can be used to exonerate the innocent and that more time is needed to gather information. Others, like San Francisco Supervisor Aaron Peskin, believe that even if AI systems achieve racial parity, facial recognition is a "uniquely dangerous and oppressive technology."

After GoPro trod on DJI's foot with its ill-fated Karma drone, DJI is stomping right back with a rugged camera of its own -- the Osmo Action. It looks like a GoPro, is similarly priced ($349 -- $51 cheaper) and pretty much goes toe-to-toe with the Hero 7 Black on key specs. Can DJI pull off what GoPro couldn't, and give its rival a run for the money? We'll get to that, but at the very least, the king of action cams has a new challenger to fend off, and that can only be a good thing for video-loving action fans. DJI's first action camera comes out fighting, with a price and feature set that should grab GoPro's attention. Neat features like a small front display and HDR video set it apart, and compatibility with its rival's accessories will make the transition easy. But there are some glaring omissions, including GPS, limited voice controls and basic social sharing options.

Good Features are the backbone of any machine learning model. And good feature creation often needs domain knowledge, creativity, and lots of time. And some other ideas to think about feature creation. TLDR; this post is about useful feature engineering methods and tricks that I have learned and end up using often. Have you read about featuretools yet? If not, then you are going to be delighted.

As a teacher of Data Science (Data Science for Internet of Things course at the University of Oxford), I am always fascinated in cross connection between concepts. To recap, Logistic regression is a binary classification method. It can be modelled as a function that can take in any number of inputs and constrain the output to be between 0 and 1. This means, we can think of Logistic Regression as a one-layer neural network. For a binary output, if the true label is y (y 0 or y 1) and y_hat is the predicted output – then y_hat represents the probability that y 1 - given inputs w and x. Therefore, the probability that y 0 given inputs w and x is (1 - y_hat), as shown below.

Large scale search advertising systems have many challenges in Natural Language Understanding and Computer Vision areas such as query and ads understanding, semantic representation, fast ads retrieval and relevance modeling, product image understanding and product detection. In his insightful talk, Bruce Zhang from Microsoft AI & Research will walk us through these various challenges and share how the Microsoft team has developed and deployed cutting-edge technologies, based on deep learning and ads domain data, in their Ads stack to improve ad quality and increase Revenue Per 1000 search (RPM). In addition, he will also share deep learning techniques used in Bing Ads such as query/ads semantic embedding models and KNN search service, query tagging model, generative models for query rewriting, DNN based query-keyword relevance model, visual product recognition models, product detection and description generation models for Product Ads. Who is this talk for? If your work touches machine learning, this talk is for you.