Daily Artificial Intelligence News Roundup #130

1. Four billion people lack an address. Machine learning could change that.

An estimated 4 billion people in the world lack a physical address. )

“As you move into a more global economy and more people order and get goods delivered at a distance, you need a more specific address than ‘the house with the red door across from the cathedral,’” says Merry Law, the president of a company that provides international addressing information. The system analyzed the density and shape of the roads to segment the network into different communities, and the densest cluster was labeled as the city center. Read More

2. Super-Intelligent AI Paperclip Maximizer Conundrum and AI Self-Driving Cars

By Lance Eliot, the AI Trends Insider

Paperclips.

Those fundamental AI-drives are aspects such as the AI having a sense of self-preservation. com/selfdrivingcars/frankenstein-and-ai-self-driving-cars/

For the potential coming singularity of AI, see my article: https://aitrends. Read More

3. How to Avoid the Potential Dangers of AI, Robots and Big Tech Companies

Even more will be done without your lifting the proverbial finger. This almost unimaginable lifestyle could become routine for the masses, given the tangible achievements of artificial intelligence (AI) and robotics to date and the low-latency-coupled-with-high-bandwidth-connectivity that 5G is on track to provide.

Despite the excitement of the likely new reality, however, AI, robots and big companies are three things a lot of people are afraid of. Read More

4. Jack Ma’s AI city, Multi-task learning in BLP, BigGan, DL cloud provider benchmark, GPU design concepts, Bengio on arms races, and more

n/a Read More

5. Artificial Intelligence Has A Probability Problem

AWS just announced Amazon SageMaker Ground Truth to help companies create training data sets for machine learning. The developer sets a 90% propensity threshold, meaning only records with a 90% probability of being accurately classified will be used as training data. Once the model is trained and deployed, it is being used on patients whose data is linked together from multiple databases using fuzzy matching on text data fields. Read More

6. Learning to Predict Depth on the Pixel 3 Phones

n/a Read More

7. Minimizing trial and error in the drug discovery process

With an estimated space of 1060 small organic molecules that could be tried and tested, it is no surprise that finding useful compounds is difficult and that the process is full of costly dead ends and surprises. Each of these fields has developed computational methods to search through molecular space and pinpoint useful leads that are followed up in the lab or in more detailed physical simulations. The abundance of data has encouraged researchers to turn to data-driven approaches to reduce the degree of trial and error in chemical development, and the aim of our paper being presented at the 2018 Conference on Neural Information Processing Systems (NeurIPS) is to investigate how recent advances, specifically in deep learning techniques, could help harness these libraries for new molecular design tasks. Read More

8. Designing a Self-Learning Tic-Tac-Toe Player

n/a Read More

9. Intriguing properties of neural networks – ShortScience.org

n/a Read More

10. How to Get a Better GAN (Almost) for Free: Introducing the Metropolis-Hastings GAN

By Ryan Turner, Jane Hung, Yunus Saatci, & Jason Yosinski

Generative Adversarial Networks (GANs) have achieved impressive feats in realistic image generation and image repair.

GANs consist of two models trained as adversaries: the generator learns the distribution of real data and the discriminator learns to distinguish generated (in other words, “fake”) samples from real data. We sample from this distribution using the Metropolis-Hastings algorithm and dub the resulting model the Metropolis-Hastings GAN (MH-GAN). Read More

11. The Future of Childcare? | What’s Next: Top Trends

n/a Read More

12. This AI-powered app aims to help people with autism improve their social skills

n/a Read More

13. Reproducing paintings that make an impression

While the original masterpieces may never be recovered, a team from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) might be able to help, with a new system aimed at designing reproductions of paintings.

“If you just reproduce the color of a painting as it looks in the gallery, it might look different in your home,” says Changil Kim, one of the authors on a new paper about the system, which will be presented at ACM SIGGRAPH Asia in December. The team found that RePaint was more than four times more accurate than state-of-the-art physical models at creating the exact color shades for different artworks. Read More

14. The real-life Robocop and other news

BBC Click’s Marc Cieslak looks at some of the best of the week’s technology news stories including;

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

15. Inside the futuristic restaurant where a robot has replaced the bartender

With architecture dating to the Middle Ages, Prague is a city whose modern identity is still defined by its many layers of ancient history, much of which remains gloriously intact.

And, yet, it is here — among a city of baroque palaces, ornate Renaissance-era chapels and cobblestone streets — that one of the first flashes of the future has suddenly, randomly emerged. Unlike other wine bars in the Czech capital, however, this one has a new twist: It is staffed by a robotic bartender who serves drinks ordered via a smartphone app, according to Reuters. Read More