Artificial Intelligence/Machine Learning Roundup #50

1. No, Robots Aren’t Taking Your Jobs… Yet – Adrian Thinnyun –

The Turing Test: what it is and what it isn’t

In 1950, Alan Turing introduced the concept of what is now known as the Turing Test in a seminal paper on artificial intelligence, Computing Machinery and Intelligence. Answering the question “Can machines think?” proved to be a messy philosophical debacle, relying too much on one’s definition of “thinking” and other subjective nuisances, so instead he devised an experiment, which goes like this:

Turing proposed that, if a computer were able to pass the Turing test reliably, it would be considered an intelligent being. Ask Duplex about its thoughts on America’s Got Talent and it won’t have a clue what to say. Read More

2. Using deep learning to predict not just what, but when

Using deep learning to predict not just what, but when

Deep learning/machine learning/AI is increasingly being used in business to predict customer behavior such as purchasing. However, none of the current methods capture the richness of customer transactions.

At BCG Gamma, we’ve recently patented (US patent number 10,002,322) a next-gen forecasting and personalization model, which we call “Crystal,” that uses deep learning to predict both what transaction will be made and when — to within a time frame of a mere few hours. Read More

3. Children and AI: ‘We have to think very carefully about the ethical boundaries’

Children and AI: ‘We have to think very carefully about the ethical boundaries’

Artificial intelligence is constantly in the news in 2018, for better or worse. Gregory and Allen agreed that there are lots of opportunities to do interesting things with AI around children’s entertainment – for example services like Netflix making sense of all the data they have on TV shows and films and on the preferences of each viewer, to make smarter recommendations. It’s all data-driven,” said Farrows. Read More

4. AI — Hype Cycle – Vidhan Singhai –

AI — Hype Cycle


All new technologies typically follow the below graph, known as the Hype Cycle. It is obviously very hard to say where we are in the above cycle this year, but I have a viewpoint on this. So in theory, it would be reasonable to say that the Trough of Disillusionment has come and gone now. Read More

5. The Future of Artificial Intelligence and what will it look like

The Future of Artificial Intelligence and what will it look like

Artificial intelligence has come a long way from being just a science fiction dream to a reality which we see today. Not long ago, holograms and smartphones were just science concepts but now the smartphones can check our health as the technology is evolving. Some of the application areas of artificial intelligence are health, education, entertainment, services, security and many other domains but these fields are specifically the most to benefit from this technology. Read More

6. Need for an artificial brain – Deepak Battini –

Need for an artificial brain

A human brain is considered as a central organ of our entire nervous system, which is connected to a spinal cord. It is made up of 100 billion neurons interconnected with each other creating a very complex network capable of learning a complex task by just looking and listening. Won’t it be wonderful if we can build an artificial form of the human brain with layers of neuron stacked and interconnected with each other and feed with data to learn and solve complex problems. Read More

7. Get ready for AI winter – Data Science at Home –

n/a Read More

8. Where did you come from? Where did you go? Where are you headed? AI will know.

Now that we’ve sketched the basic model for describing a user’s experience (again, we’re discussing this in the context of a website visit, but the experience could be anything — and for that matter, so could the user), let’s consider how we might build a data set consisting of many users’ visits to a website.

To start, we’re going to want to define the major actions (the Nodes) for the machine.

When the user visits the website, we’ll record all of their actions: from Pips to Nodes, Nodes to Chains, and finally to (or not to) an Endpoint. Read More

9. An Introduction to Biomedical Image Analysis with TensorFlow and DLTK

An Introduction to Biomedical Image Analysis with TensorFlow and DLTK

By Martin Rajchl, S. Ira Ktena and Nick Pawlowski — Imperial College London

DLTK, the Deep Learning Toolkit for Medical Imaging extends TensorFlow to enable deep learning on biomedical images.

This blog post serves as a quick introduction to deep learning with biomedical images, where we will demonstrate a few issues and solutions to current engineering problems and show you how to get up and running with a prototype for your problem. Read More

10. This ‘smart’ prosthetic ankle could change the lives of amputees

n/a Read More

11. The DOD’s App Store Does This One Crucial Thing to Stay Secure

Every day, companies like Google and Apple wage a constant battle to keep malicious apps out of their marketplaces and off people’s phones. NGA is a combat support organization that primarily assesses and distributes geospatial intelligence.

“We recognized that we did not know everything when it came to apps, and we wanted to be using the innovation that was happening in the commercial sector,” says Joedy Saffel, division chief and source director of NGA who has worked on the GEOINT App Store from the beginning. Read More

12. ‘China’s Google’ releases its first AI chip

n/a Read More

13. DeepMind’s AI agents are better than humans at being your teammate

DeepMind has developed a way for AI programs to learn how to team up effectively in a simple video game. Most strikingly, the AI agents can also collaborate with human players—and those players say the programs are better teammates than most people. Teamwork is extremely difficult to develop effectively in AI programs, because it involves dealing with a complex and ever-changing situation. Read More

14. Time to stop worrying about robots taking our jobs and start dealing with it

A new report out from a think tank called the Center for Global Development argues that we need to shift our attention from guessing how many jobs will be eliminated to trying to fix the issue.

Some background: A slew of research projects have made a wide variety of predictions about automation-caused job loss. Predictions differ by tens of millions of jobs, even when comparing similar time frames. Read More

15. This robot is not throwing away its shot

Researchers have used a technique called one-shot learning to teach a robot to pick up things it’s never seen before. It’s AI software that can perform a task after being given a single data point.

The news: In a paper published on arXiv (PDF), researchers at UC Berkeley revealed they had developed a robotic system that can pick up an object it’s seeing for the first time. Read More

16. AI can now ‘listen’ to machines to tell if they’re breaking down

Sound is everywhere, even when you can’t hear it.

It is this noiseless sound, though, that says a lot about how machines function. Both companies were also finalists in last year’s New Energy Challenge, an initiative by Shell, YES!Delft, and Rockstart that looks at innovative technologies and solutions within European and Israeli startups for the energy transition. Read More

17. DARPA using AI to find better chemistry for batteries and bombs

DARPA is developing software tools based on machine learning and expert-encoded rules to recommend chemical synthesis recipes to optimize cost, time, safety, or waste reduction.

Make-It is achieving success due to its multidisciplinary focus with a talented group of researchers, combining computer science, organic chemistry, and chemical engineering expertise to address three areas: automated molecule design; automated synthesis, or production; and rapid reaction screening.

The University of Glasgow is developing software to design and 3-D print small, portable reactors to provide manufacturing of specific molecules “on-the-go. Read More

18. Google’s DeepMind taught AI teamwork by playing Quake III Arena

Google’s DeepMind today shared the results of research and experiments in which multiple AI systems were trained to play Capture the Flag on Quake III Arena, a multiplayer first-person shooter game. An AI trained in the process is now better than most human players in the game, regardless of whether it’s playing with a human or machine teammate.

The AI, named For the Win (FTW), played nearly 450,000 games of Quake III Arena to gain its dominance over human players and establish its understanding of how to effectively work with other machines and humans. Read More

19. Self-driving cars are headed toward an AI roadblock

If you believe the CEOs, a fully autonomous car could be only months away. There’s growing concern among AI experts that it may be years, if not decades, before self-driving systems can reliably avoid accidents. Over the past ten years, deep learning — a method that uses layered machine-learning algorithms to extract structured information from massive data sets — has driven almost unthinkable progress in AI and the tech industry. Read More

20. Facial recognition’s moral dilemma

Tech companies that focus on AI-powered computer vision “are struggling to balance business opportunities with difficult moral decisions that could turn off customers or their own workers,” the AP’s Matt O’Brien reports.

Why it matters: It’s part of a growing wave of concern about how Artificial Intelligence technology is used, especially involving facial recognition, Axios’ Ina Fried emails. Read More

21. US town swaps fireworks show for drones

The US town of Aspen is swapping its annual Fourth of July fireworks display for a drone show – because of a wildfire risk in the drought-hit area.

Officials say 50 drones will light up the night sky in patriotic colours to mark Independence Day.

This comes after Colorado’s fire authorities introduced strict fire restrictions across huge swathes of the western state. Read More

22. Self-drive buses enter ‘mass production’

One of China’s biggest technology companies has declared it has begun mass production of a self-driving bus.

The classification – set by the transport engineering body SAE International – refers to highly automated driving systems that can cope with most driving conditions, even if a human fails to respond appropriately to a request to intervene.

It has no driver’s seat, steering wheel or pedals. Read More

23. Asimo Still Improving Its Hopping and Jogging Skills

We learned last week that Honda is putting Asimo out to pasture, so to speak, which is a little sad, but not too sad: Honda is doing this because they want to instead focus on the other, more useful humanoid robots that they’ve been working on recently, like E2-DR. Honda learned a lot about humanoid robotics from Asimo, and even though we haven’t seen Asimo do much in the way of new stuff over the past few years, the robot has still been under active development.

Nobody likes to see robots getting pushed or kicked, but we can make exceptions when roboticists are doing it for research-related reasons—for example, to demonstrate how resilient their quadrupeds or humanoids are by trying to shove them over. Read More

24. Isn’t it time we declared our independence from bloatware?

So you’ve just bought the best Windows laptop, you’ve gritted your teeth through Cortana’s obnoxiously cheery setup narration, and the above screenshot is the Start menu you’re presented with.
Before anyone assumes that this is just a rant against and about Windows, I’ll happily include Apple’s iOS and some varieties of Google’s Android in my scorn. We can have debates about whether pushy notifications about ancillary services from the OS maker necessarily constitute bloat, but there’s little room for disagreement when it comes to third-party additions. Read More

25. Keeping America first in quantum computing means avoiding these five big mistakes

After watching China overtake its lead in artificial intelligence, the US is determined to keep the global top spot in quantum technology.

Two separate pieces of legislation being floated in Congress would boost federal spending on quantum research and encourage more public-private partnerships in the field. The European Union has launched a multi-year initiative backed by an investment of around $1. Read More

26. Facebook confirms that it’s acquiring Bloomsbury AI

Facebook announced this morning that the London-based team at Bloomsbury AI will be joining the company.

My colleague Steve O’Hear broke the news about the acquisition, reporting that Facebook would deploy the team and technology to assist in its efforts to fight fake news and address other content issues. Instead, it says the team’s “expertise will strengthen Facebook’s efforts in natural language processing research, and help us further understand natural language and its applications” — but it certainly seems possible that those applications could include detecting misinformation and other problematic content. Read More

27. How Google and Facebook Are Monopolizing Ideas

In early May Google banned bail-bond companies from advertising on its platforms. They use “opaque financing offers that can keep people in debt for months or years. ”

That Google can ban ads from an industry that offends its values is not, by itself, noteworthy. Read More

28. Opinion: Data is Holding Back AI

By Sultan Meghji, Founder & Managing Director at Virtova

I remember grumbling, “Good lord this is a waste of time,” in 1992 while I was working on an AI application for lip-reading. The grumble escaped my lips because I felt like I was spending half my time inputting data cleanly into the video processing neural network.

To be sure, there are significant improvements being made to decrease the amount of information needed to train AI systems and in building effective learning transference mechanisms. Read More

29. 13 Experts Weigh in on What We Should Teach Robots

We’ve reached a critical point in the development of robots. ” – George Konidaris

Teaching a machine to live alongside humans means that nothing can be taken for granted, no matter how obvious a certain set of information might seem. This research could be key in creating robots that can complete a variety of tasks, without the requirement of specific programming to facilitate each task. Read More

30. This AI analyzes ash to figure out the cause of a volcanic eruption


Now, scientists from the Earth-Life Science Institute at the Tokyo Institute of Technology have developed an artificial intelligence (AI) program that analyzes volcanic ash particles to determine their shape. The Tokyo team’s AI is what’s known as a convolutional neural network (CNN), a kind of AI frequently used to analyze images. Read More

31. Facial recognition tech is no match for Juggalo makeup

LOOKING FOR A REASON TO BECOME A JUGGALO?  Twitter user @tahkion has discovered a new way to beat facial recognition systems, computer programs that analyze images or videos of people’s faces for the purposes of identifying them. In a series of tweets on Saturday, the computer science blogger for WonderHowTo described how the face makeup favored by Juggalos — die-hard fans of the musical group Insane Clown Posse — confuses the systems to the point they can’t accurately identify people. As facial recognition systems become more sophisticated, they’re also becoming more commonly used. Read More

32. Crop-counting robot

To succeed, they must locate genes for high-yielding, hardy traits in crop plants’ DNA. A robot developed by the University of Illinois to find these proverbial needles in the haystack was recognized by the best systems paper award at Robotics: Science and Systems, the preeminent robotics conference held last week in Pittsburgh. ”

Crop breeders run massive experiments comparing thousands of different cultivars, or varieties, of crops over hundreds of acres and measure key traits, like plant emergence or height, by hand. Read More

33. [1807.00818] Improving part-of-speech tagging via multi-task learning and character-level word representations

n/a Read More

34. [1807.01081] Solving Atari Games Using Fractals And Entropy

n/a Read More

35. Predicting Future Online Threats with Big Data

Cyber crime and online threats are on the rise, a rapid rise. To counter this rapid increase, new methods are also on the rise that employ inventive and sophisticated methods to detect, prevent and predict future online threats. The main driver behind these new methods is big data. Read More

36. Learning Montezuma’s Revenge from a Single Demonstration

We’ve trained an agent to achieve a high score of 74,500 on Montezuma’s Revenge from a single human demonstration, better than any previously published result. Our algorithm is simple: the agent plays a sequence of games starting from carefully chosen states from the demonstration, and learns from them by optimizing the game score using PPO, the same reinforcement learning algorithm that underpins OpenAI Five.

Why Exploration is Difficult

Model-free RL methods like policy gradients and Q-learning explore by taking actions randomly. Read More

37. The New Age of Intelligent Machines – Prof. Shimon Ullman,

n/a Read More

38. Scientists have built a robotic ‘dragon’

n/a Read More

39. Relativistic GAN

I just released my latest paper called The relativistic discriminator: a key element missing from standard GAN (code to implement relativistic GANs here). In this paper, I argue that standard GAN (SGAN) is missing a fundamental property, i. I was able to train relativistic SGAN and Least squares GAN (LSGAN) on a small sample of N=2011 with 256×256 pictures, which is something that SGAN, LSGAN cannot even do (they get stuck at generating noise) and that Spectral GAN and WGAN-GP do poorly. Read More