Machine Learning Leads Mathematicians to Unsolvable Problem

In exploring a machine learning problem, a team of researchers at the Technion-Israel Institute of Technology in Haifa discovered a mathematically unanswerable question associated with logical paradoxes defined by mathematician Kurt Godel. The team determined “learnability”—whether an algorithm can extract a pattern from limited data—is connected to Godel’s continuum hypothesis, in which a statement can be neither validated nor invalidated via standard mathematical language. This determination stemmed from an investigation into the link between learnability and “compression,” which entails finding a way of abstracting the salient features of a large dataset in a smaller dataset. The team described learnability as the ability to make predictions about a large dataset by sampling a small number of data points. Although the validation of the continuum hypothesis means a small sample is sufficient to make the extrapolation, its invalidation means no finite sample can ever be sufficient.

More info here: Nature, Davide Castelvecchi

Anuncios
Publicado en Ciencia y programación

A Fleet of Autonomous Robots Is Making One of the World’s Oldest Foods

An autonomous, bread-making robotic oven made by Wilkinson Baking of Walla Walla, WA, can produce 235 loaves daily, ranging from white and whole wheat to nine grain to honey oat and rye. The BreadBot mixes, kneads, bakes, and cools the dough without human assistance. Once the BreadBot mixes the dough into balls, it transfers them onto a conveyor belt, shapes them, and places them inside individual trays for baking; afterwards, a robotic arm moves the loaf to a vending machine for purchase via touchscreen. More than 70 sensors monitor the bread 100 times a second and adjust the baking process in real time. Wilkinson Baking’s Randall Wilkinson said, “You’ve got a lot of [artificial intelligence] data that you can crunch for optimum performance, and all of these machines are Internet-connected.”

More info here: The Washington Post, Peter Holley

Publicado en Ciencia y programación

Virtual Reality Gets Real In the Operating Room

A growing number of hospitals and medical centers are embracing virtual technology (VR), with the goal of providing better and faster training for resident doctors and surgeons. Stanford University students learn anatomy by walking around a lifelike digital hologram of a lung, and transport themselves inside a heart to see the valves and pumping blood. VR technology helps students learn faster, which is especially important in countries like China and India, where a combined 6 million new physicians will be needed by next year. VR can be used either as a fully immersive experience, in which users see only a computer-generated environment, or as a part of mixed reality, in which three-dimensional images are projected onto the physical world. The University of Washington in Seattle’s Richard Satava said VR “gives us a way to judge whether the medical student has learned what they are supposed to learn.”

More info here: Fortune, Andrew Zaleski

Publicado en Ciencia y programación

One Giant Step for a Chess-Playing Machine

 

The success of DeepMind’s chess-playing AlphaZero algorithm mirrors the evolution of machine learning and its ability to deduce rules via experiential learning. AlphaZero’s ability to express insight also is unsettling, as it played gambits and took risks in matches against both human grandmasters and other champion game-playing programs. For example, in tournaments against the Stockfish algorithm, AlphaZero won consistently by thinking smarter instead of faster, analyzing only 60,000 positions a second versus Stockfish’s 60 million. World chess champion Garry Kasparov said AlphaZero’s superiority was rooted in a style of gameplay that “reflects the truth” about the game, rather than “the priorities and prejudices of programmers.” Innate in this milestone is the potential for insightful algorithms to tackle more challenging problems that could benefit human knowledge and society; however, with this possibility comes the risk of such algorithms ultimately overtaking human insight.

More info here: The New York Times, Steven Strogatz

Publicado en Ciencia y programación

The Yoda of Silicon Valley

Stanford computer scientist Donald Knuth has been known as one of the leaders in the field of computing for 50 years, having written “The Art of Computer Programming” in 1968, which has served as the bible of its field with more than 1 million copies in print. Knuth is known for introducing the notion of “literate programming,” emphasizing the importance of writing code that is readable by humans as well as computers. In addition, Knuth helped develop some of computing’s most important algorithms, such as the Knuth-Morris-Pratt string-search algorithm, which finds all occurrences of a given word or pattern of letters in a text. Knuth’s work opened the door to algorithms so complicated they cannot be read, and even algorithms written by other algorithms. Said Knuth: “I am worried that algorithms are getting too prominent in the world. It started out that computer scientists were worried nobody was listening to us. Now I’m worried that too many people are listening.”

More info here: The New York Times, Siobhan Roberts

Publicado en Ciencia y programación

Israel Joins the Race to Become a Quantum Superpower

Israel’s Defense Ministry recently announced an investment of 100 million shekels (more than $26 million) in a research fund for quantum computing to enhance the nation’s intelligence-gathering capacity. Liat Maoz with Israel’s Council for Higher Education said quantum computers could crack complex encryption codes, as well as enable completely secure communications. Said the Council for Higher Education’s Yaffa Zilbershatz, “The global race is already underway and various countries are investing huge sums in developing the field—and if we do not run forward, the State of Israel will be left behind.” Tamir Libel with the Institut Barcelona d’Estudis Internacionals in Spain said quantum computing could be used as an alternative method for more accurate synchronization between military navigation systems.

More info here: The Jerusalem Post, Anna Ahronheim; Maayan Hoffman

Publicado en Ciencia y programación

White House Outlines Five-Year STEM Push

The Trump administration announced a five-year strategic plan for science, technology, engineering, and math (STEM) education, issuing a call to action for nationwide collaboration with students, families, educators, communities, and employers. The administration’s goal has three facets: for every American to master basic STEM concepts in order to respond to technological change; to increase access to STEM among historically underserved students; and to encourage students to pursue STEM careers. Achieving these goals requires strengthening partnerships between schools, businesses, nonprofits, and others in order to leverage resources and expertise in the STEM field, according to the White House. The plan also urges educators to make STEM “more meaningful and inspiring” through strategies like project-based learning, science fairs, robotics clubs, invention challenges, and gaming workshops. One of the major challenges White House officials see in fulfilling the mission is a lack of STEM teachers in grades K-12.

More info here: U.S. News & World Report, Lauren Camera

Publicado en Ciencia y programación