Lifeboat Foundation

Learning how to learn is something most humans do well, by leveraging previous experiences to inform the learning processes for new tasks. Endowing AI systems with such abilities however remains challenging, as it requires the machine learners to learn update rules, which typically have been manually tuned for each task.

The field of meta-learning studies how to enable machine learners to learn how to learn, and is a critical research area for improving the efficiency of AI agents. One of the approaches is for learners to learn an update rule by applying it on previous steps and then evaluating the corresponding performance.

To fully unlock the potential of meta-learning, it is necessary to overcome both the meta-optimization problem and myopic meta objectives. To tackle these issues, a research team from DeepMind has proposed an algorithm designed to enable meta-learners to teach themselves.

Two interactive web experiences let you explore the Martian surface, as seen by cameras aboard the rover and orbiters flying overhead.

It’s the next best thing to being on Mars.

Continue reading “Interactive Web Experiences: Take a 3D Spin on Mars and Track NASA’s Perseverance Rover” »

Even the most highly trained and experienced person sometimes needs a hand.

https://www.youtube.com/watch?v=4DMfvQbP8cE&feature=share

This film looks like an unexpectedly positive, interesting take on post-apocalyptic AI and its creator (or creators) than we are usually presented with from Hollywood. Between this and the new Foundation adaptation (that comes out in less than a WEEK!), maybe it’s time to forgive them for their many sins and finally subscribe to Apples streaming service. Or maybe not. 🤔

First trailer for Finch starring Tom hanks.

Interesting.

New multi-mission unmanned will support ground forces in the field while minimizing threats to human lives.

The American West is running out of water—and Big Oil, of all things, can help fix it.

It’s time to repurpose our oil and gas infrastructure to do something good for the planet: transport water to the parched West.

Ancient DNA extracted from human bones has rewritten early Japanese history by underlining that modern day populations in Japan have a tripartite genetic origin—a finding that refines previously accepted views of a dual genomic ancestry.

Twelve newly sequenced ancient Japanese genomes show that modern day populations do indeed show the genetic signatures of early indigenous Jomon hunter-gatherer-fishers and immigrant Yayoi farmers—but also add a third genetic component that is linked to the Kofun peoples, whose culture spread in Japan between the 3^rd and 7^th centuries.

“We were confident that the rocks of the Champua region were even older than previously thought,” says Mazumder, 53 now an associate professor of applied geosciences at the German University of Technology in Oman.

They were right. In 2,018 they published a paper in the journal Scientific Reports on two zircon crystals they extracted from rocks taken from the Champua site. While the rocks were 3.4 billion years old, the crystals were much older, at an estimated 4 and 4.2 billion years old.

Altana AI, a startup building a database for global supply chain networks, today announced that it raised $15 million in a series A funding round led by GV with participation from Floating Point, Ridgeline Partners, Amadeus Capital Partners, and Schematic Ventures. The proceeds, which bring the company’s total raised to $22 million to date, will be used to further develop Altana’s data and AI systems and launch new machine learning and network analysis tools, according to CEO Evan Smith.

Trade wars, the rise of ecommerce, pandemic supply chain shocks, and sustainability concerns are driving fundamental changes to supply chain networks and global trade flows. Nearly 75% of companies report supply chain disruptions in some capacity due to pandemic-related transportation restrictions. And in a recent IBM survey, 40% of executives stressed the need for spare capacity to weather future crises.

Altana’s product aims to solve these challenges with a platform that connects and learns from billions of supply chain data points. It answers questions about products, shipments, companies, and networks, filtering out illicit trade and targeting bad actors and security threats across global commerce networks.