Artificial Intelligence/Machine Learning Roundup #54

1. Did A Robot Write This Blog Post?

traffic last week on the way to a client meeting when I caught an interesting story on 1A radio about “Artificial Intelligence.

My ears initially perked when I heard about the role of artificial intelligence (AI) in the royal wedding coverage transforming some of today’s major newsrooms.

I really encourage you to listen to the whole program because this short post alone really can’t do justice to the range of interesting topics the guests covered — including the future of storytelling, reporting ethics, the newsroom of the future and more. Read More

2. Interpreting AI Compute Trends – AI Impacts

Over the last few years, we know that AI experiments have used much more computation than previously.

The rate of growth of experiments according to this AI-Compute trend is astoundingly fast, and this deserves some analysis. 5 more years, the amount of compute used in an AI experiment will have passed some interesting milestones. Read More

3. Artificial intelligence helps researchers predict drug combinations’ side effects

Last month alone, 23 percent of Americans took two or more prescription drugs, according to one CDC estimate, and 39 percent over age 65 take five or more, a number that’s increased three-fold in the last several decades. And if that isn’t surprising enough, try this one: in many cases, doctors have no idea what side effects might arise from adding another drug to a patient’s personal pharmacy. That system, called Decagon, could help doctors make better decisions about which drugs to describe and help researchers find better combinations of drugs to treat complex diseases. Read More

4. Discover Financial Services applies machine learning through a Robocar event powered by Amazon SageMaker

The Discover Financial Services (DFS) team members who attended AWS re:Invent agreed that the Robocar Rally was an extremely impactful experience. By participating in this hackathon, six members of Discover’s core team received hands-on experience using machine learning (ML) and deep learning on AWS. Discover’s Cloud Center of Excellence (CCoE) later proposed an idea: Let’s find a way to replicate a similar event onsite at the Discover Headquarters. Read More

5. How to Stop Artificial Intelligence Being Biased

ACM TechNews

How to Stop Artificial Intelligence Being Biased

Niki Kilbertus and colleagues at the Max Planck Institute for Intelligent Systems in Germany have developed a new method to avoid embedding bias into machine-learning algorithms.

Their technique involves incorporating sensitive data in the training process while including an independent regulator and applying encryption mathematics.

When training the artificial intelligence (AI), an organization can use as much non-sensitive data as required, but both the organization and the regulator only receive sensitive data in an encrypted form. Read More

6. [1807.03586] Difficulty Controllable Question Generation for Reading Comprehension

n/a Read More

7. Deep Learning Weekly – 🤖 – Issue #89: Fairness, Driving in a Day, Fake SloMo, L2 Regularization, RNNs, Art, Prototyping with TF, Edge Performance and more…

Hey and Welcome to a new week in deep learning.
As always, we hope you’ll enjoy reading as much as we did and would appreciate you sharing this newsletter with friends and colleagues.

See you next week. Read More

8. Boston Dynamics’ SpotMini Could Deliver Your Next Package

n/a Read More

9. IBM Research AI at IJCAI-ECAI 2018

n/a Read More

10. Multi-Face Tracking to Help AI Follow the Action

At the recent 2018 Conference on Computer Vision and Pattern Recognition, I presented a new algorithm for multi-face tracking, an essential component in understanding video.

A method for multi-face tracking

Collaborating with Professor Ying Hung, of Department of Statistics and Biostatistics in Rutgers University, we developed a method to spot different individuals in a video sequence and to recognize them if they leave then re-enter the video, even if they look very different.

Multi-face tracking in music videos

To evaluate the performance of our approach, we compared it against state-of-the-art methods in analyzing challenging datasets of unconstrained videos. Read More

11. Best of for AI, Machine Learning, and Deep Learning – June 2018

n/a Read More

12. The story of an Office AI feature: How AI can promote efficient meeting preparation

He was spending so much time searching through old emails and attachments to prepare for meetings: Could AI help him be better prepared for meetings.

At the same time, Paul Bennett’s Information and Data Sciences team in Microsoft Research AI was in the second year of a project to help people find the right information, effortlessly.

Introducing CAPERS

The feature is called a Calendar-Aware Proactive Email Recommender System (CAPERS). Read More

13. BRUNO: A Deep Recurrent Model for Exchangeable Data on

n/a Read More

14. An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution

Uber uses convolutional neural networks in many domains that could potentially involve coordinate transforms, from designing self-driving vehicles to automating street sign detection to build maps and maximizing the efficiency of spatial movements in the Uber Marketplace. Since such architectures are widespread, we should expect that they excel at simple tasks like painting a single pixel in a tiny image, right. In our paper, An Intriguing Failing of Convolutional Neural Networks and the CoordConv Solution, we expose and analyze a generic inability of convolutional neural networks (CNNs) to transform spatial representations between two different types: coordinates in (i, j) Cartesian space and coordinates in one-hot pixel space. Read More

15. ‘Experience what characters are feeling’: U of T researchers use AI to add ‘4D’ effects to movies

While a small number of so-called “4D” movies that add a physical element already exist, researchers from the University of Toronto are working on a way to apply the feature more broadly.

“Usually the chair will shake, there can be splashing or some other kind of interaction while watching the film,” says Yuhao Zhou, a fourth-year undergraduate in the Edward S. “Right now all these effects are created from the first phase of production. Read More

16. The US may have just pulled even with China in the race to build supercomputing’s next big thing

There was much celebrating in America last month when the US Department of Energy unveiled Summit, the world’s fastest supercomputer. Now the race is on to achieve the next significant milestone in processing power: exascale computing.

This involves building a machine within the next few years that’s capable of a billion billion calculations per second, or one exaflop, which would make it five times faster than Summit (see chart). Read More

17. A replacement for traffic lights gets its first test

The world’s first traffic light system began operating near the Houses of Parliament in London in 1868. It consisted of a set of gas lights operated by a policeman and designed to control the flow of horse-drawn traffic across the Thames. These guys have tested a way of ridding our streets of traffic lights entirely and replacing them with a virtual system instead, saying that their system has the potential to dramatically reduce commuting times. Read More

18. How AI is transforming tennis

Click goes to Wimbledon to see how artificial intelligence (AI) is being used to highlight the best moments in the tennis tournament and at the same time help players improve their game.

See more at Click’s website and @BBCClick. Read More

19. Hackers can purchase government login credentials for cheap on the dark web

McAfee’s Advanced Threat Research team recently discovered that hackers have access to many organizations that have weak credentials when using Microsoft’s Remote Desktop component in Windows-based systems.

Microsoft’s Remote Desktop Protocol (RDP) essentially allows you to connect and use a Windows-based PC from a remote location. McAfee found connections up for sale across various RDP shops on the dark web ranging between a mere 15 to a staggering 40,000 connections. Read More

20. Toyota AI Ventures Wants to Fund Your Mobile Manipulation Startup

Today, Toyota AI Ventures in partnership with the Toyota Research Institute (TRI) is opening a “call for innovation” with the goal of throwing money at anyone with good ideas for “improving mobile manipulation technology for assistive robots that can help people in and around the home. ”

Launched last year with US $100 million to spread around, Toyota AI Ventures has already started funding robotics companies, including Intuition Robotics and Realtime Robotics. Software solutions could include ways to compensate for lower-precision, lower-fidelity hardware; algorithms to learn from or annotate data; and ways to apply lessons learned from simulation. Read More

21. Here’s What Sophia, the First Robot Citizen, Thinks About Gender and Consciousness

In a video that’s as unsettling as it is awe-inspiring, Sophia — the world’s first robot citizen — breaks down everything from gender to ethical robot design.

Since Sophia was activated in April 2015, she has appeared publicly to speak about women’s rights issues, her own citizenship and other topics. ” [Machine Dreams: 22 Human-Like Androids from Sci-Fi]

At Brain Bar, Sophia had a lot to say. Read More

22. What If AI Psychopath ‘Norman’ and Amazon’s Alexa Got Together?

If, like Rip Van Winkle, you’ve been asleep for the last decade and have just woken up, your flip phone has become super-popular among retro technologists and survivalists alike, and, oh yeah, artificial intelligence (AI) is either going to kill you or save you.

AI is the latest in a long line of technology buzzwords that have gripped society, and if we are to believe the people at the respected technology analysts firm Gartner Inc. As unnerving as the surreal robotics being cooked up at Boston Dynamics or the deployment of facial recognition AI in Chinese public schools may seem, this technology is a product of the human condition and as such, we are embedding our own culture within its coded DNA. Read More

23. Researchers develop accent detection AI to improve speech recognition

Few things are more frustrating than a speech recognition system that doesn’t understand your accent. Training a model requires a wealth of data, and some dialects are less common than others. ”

The team sourced its data from the Carnegie Mellon University (CMU) Pronouncing Dictionary, which contains thousands of audio recordings of English speakers reading common words. Read More

24. Faculty, Students Aim To Drive Autonomous Vehicle Policies for Pittsburgh – News – Carnegie Mellon University

July 09, 2018

Faculty, Students Aim To Drive Autonomous Vehicle Policies for Pittsburgh

Carnegie Mellon University students and faculty recently met with Pittsburgh civic leaders to discuss new policy recommendations for autonomous vehicles.

A weekend course in February hosted by the Heinz College of Information Systems and Public Policy, assembled students and faculty from Heinz College, the College of Engineering, College of Fine Arts and Dietrich College of Humanities and Social Sciences and expert guests with the goal of creating collaborative and comprehensive policy recommendations.

“We’re already seeing a battle between parking, loading, accessibility, freight, and transit in downtown and the CBD,” Luther said. Read More

25. The Coolest Uses of AI Right Now – Mat Cole –

The Coolest Uses of AI Right Now

The future is here.


Sentient robots creating meaningful art is the stuff of sci-fi, but that doesn’t mean AI can’t help humans tap into deeper creativity. A machine could craft a tune, for instance, by “listening” to music and producing similar sounds. Read More

26. OpenAI Glow Helps AI Agents Learn with Small Datasets

OpenAI Glow Helps AI Agents Learn with Small Datasets

Since the early days of machine learning, artificial intelligence scenarios have faced with two big challenges in order to experience mainstream adoption. First, we have the data efficiency problem that requires machine or deep learning models to be trained using large and accurate datasets which, as we know, are really expensive to build and maintain.

Generative models are one of the deep learning disciplines that focuses on addressing the two challenges mentioned above. Read More

27. Demystifying artificial intelligence, machine learning and use of statistics

Demystifying artificial intelligence, machine learning and use of statistics

In an AI first world, it’s extremely important for businesses looking to tap the power of AI, to be able to clearly understand the wholeness of data science, what AI encompasses, how powerful is ML and when statistics comes into play.

Sometimes it’s the hardest to fathom what the human brain can do and articulate human intelligence. An Artificially Intelligent machine, in short mimics human intelligence — constantly observing the environment, interpreting information, predicting outcomes, making decisions and taking actions. Read More