Archive for the ‘Computer science’ Category

Google’s AI group moves on from Go, tackles Quake III Arena

May 30th, 2019
Representations of bots on a Quake map.

Enlarge / Representation of some of the behaviors developed by the FTW algorithm. (credit: Deep Mind)

Google's AI subsidiary Deep Mind has built its reputation by building systems that learn to play games by playing each other, starting with little more than the rules and what constitutes a win. That Darwinian approach of improvement through competition has allowed Deep Mind to tackle complex games like chess and Go, where there are vast numbers of potential moves to consider.

But at least for board games like those, the potential moves are discrete and don't require real-time decisionmaking. It wasn't unreasonable to question whether the same approach would work for completely different classes of games. Such questions, however, seem to be answered by a report in today's issue of Science, where Deep Mind reveals the development of an AI system that has taught itself to play Quake III Arena and can consistently beat human opponents in capture-the-flag games.

Not a lot of rules

Chess' complexity is built from an apparently simple set of rules: an 8 x 8 grid of squares and pieces that can only move in very specific ways. Quake III Arena, to an extent, gets rid of the grid. In capture-the-flag mode, both sides start in a spawn area and have a flag to defend. You score points by capturing the opponent's flag. You can also gain tactical advantage by "tagging" (read "shooting") your opponents, which, after a delay, sends them back to their spawn.

Read 15 remaining paragraphs | Comments

Posted in AI, Computer science, Deep Mind, gaming, quake III, science | Comments (0)

US computer science grads outperforming those in other key nations

March 23rd, 2019
A chocolate cake is decorated by the plastic figurine of a celebratory graduate, complete with diploma and mortarboard.

Enlarge (credit: David Goehring / Flickr)

There's a steady flow of reports regarding the failures of the US education system. Read the right things and you'll come away convinced that early grades fail to teach basic skills, later grades fail to prepare students for college, and colleges students fail so much that they can't cope with the world outside the campus walls. But this week brought a bit of good news for one particular area: college-level computer science programs appear to be graduating some very competitive students.

This comes despite the fact that US students enter colleges behind their peers in other countries.

The work, done by an international team of researchers, compares US college seniors to those of three countries where US companies have outsourced some of their work: China, India, and Russia. All of these countries have a reputation for first-rate computing talent, with India and China developing large internal markets as well. Many students from these countries also come to study in the US, while Russia and China have been involved in cyber attacks against the United States and/or companies based here.

Read 8 remaining paragraphs | Comments

Posted in Computer science, education, science | Comments (0)

Information overload study we covered has been retracted

January 10th, 2019
Sorry, I’m not home right now.

Enlarge / Sorry, I’m not home right now. (credit: flickr user: Rosmarie Voegtli)

January 10, 2019: In 2017, we covered a study that suggested information overload may be responsible for the viral spread of faulty information. The study was based on a mix of modeling of artificial "agents" that forwarded information to their peers, and real-world data obtained from Twitter. In attempting to follow up on their own work, the researchers who produced it discovered two problems: a software bug in their analysis pipeline, and a graph that was produced using invalid data.

Combined, these suggest the model they favored—that high- and low-quality information were equally likely to spread—wasn't valid. While this doesn't alter the empirical data they obtained, it does influence their analysis of it, so they have chosen to retract the paper.

The retraction highlights one of the frequently overlooked aspects of scientific reproducibility. Problems with published work are frequently identified not by repeating the exact same experiments, but by attempts to build or expand upon them.

The original story follows. Credit to Retraction Watch for identifying the retraction.

Original story follows

Once upon a time, it wasn’t crazy to think that social media would allow great ideas and high-quality information to float to the top while the dross would be drowned in the noise. After all, when you share something, you presumably do so because you think it’s good. Everybody else probably thinks what they’re sharing is good, too, even if their idea of “good” is different. But it’s obvious that poor-quality information ends up being extremely popular. Why?

That popularity might be a product of people’s natural limitations: in the face of a flood of information and finite attention, poor quality discrimination ends up being a virtual certainty. That’s what a simulation of social media suggests, at least.

Read 16 remaining paragraphs | Comments

Posted in Behavioral science, Computer science, science, social media | Comments (0)

Move over AlphaGo: AlphaZero taught itself to play three different games

December 6th, 2018
Starting from random play and knowing just the game rules, AlphaZero defeated a world champion program in the games of Go, chess, and shoji (Japanese chess).

Enlarge / Starting from random play and knowing just the game rules, AlphaZero defeated a world champion program in the games of Go, chess, and shoji (Japanese chess). (credit: DeepMind Technologies, Ltd.)

Google's DeepMind—the group that brought you the champion game-playing AIs AlphaGo and AlphaGoZero—is back with a new, improved, and more-generalized version. Dubbed AlphaZero, this program taught itself to play three different board games (chess, Go, and shoji, a Japanese form of chess) in just three days, with no human intervention.

A paper describing the achievement was just published in Science. "Starting from totally random play, AlphaZero gradually learns what good play looks like and forms its own evaluations about the game," said Demis Hassabis, CEO and co-founder of DeepMind. "In that sense, it is free from the constraints of the way humans think about the game."

Chess has long been an ideal testing ground for game-playing computers and the development of AI. The very first chess computer program was written in the 1950s at Los Alamos National Laboratory, and in the late 1960s, Richard D. Greenblatt's Mac Hack IV program was the first to play in a human chess tournament—and to win against a human in tournament play. Many other computer chess programs followed, each a little better than the last, until IBM's Deep Blue computer defeated chess grand master Garry Kasparov in May 1997.

Read 11 remaining paragraphs | Comments

Posted in AI, alphago, AlphaZero, Artificial intelligence, Computer science, deep learning, deepmind, game theory, gaming, Gaming & Culture, neural networks, reinforcement learning, science | Comments (0)

New software will let artists control how light interacts with objects

November 30th, 2018
Clouds contain billions of individual water droplets that are difficult to plot in computer graphics for movie scenes.

Enlarge / Clouds contain billions of individual water droplets that are difficult to plot in computer graphics for movie scenes. (credit: Dartmouth Visual Computing Lab)

Animators will now be able to precisely control how microscopic particles interact with light in their renderings of objects, thanks to a research collaboration between computer scientists at Dartmouth University and staff scientists at Pixar and Disney. The team will describe this new work next week at the SIGGRAPH Asia event in Tokyo, Japan; a paper is also forthcoming in the journal Transactions on Graphics.

The breakthrough will allow animation artists more creative leeway when designing the look of various objects by giving them the ability to customize the way light travels through them. It should have the biggest impact on renderings of so-called "volumetric materials"—clouds, fog, mist, skin, or marble statues, for instance. (Marble is a material that reflects some light off the surface but allows some to pass through, giving it a translucent appearance.)

"There is a whole range of dramatically different appearances that artists just couldn't explore until now," said Dartmouth co-author Wojciech Jarosz. "Previously, artists basically had one control that could affect the appearance of a cloud. Now it's possible to explore a vastly richer palette of possibilities, a change that is as dynamic as the transition from black-and-white images to color."

Read 7 remaining paragraphs | Comments

Posted in biomimicry, Computer science, Gaming & Culture, mathematical modeling, moviemaking, Physics, science, siggraph | Comments (0)

AIs trained to help with sepsis treatment, fracture diagnosis

October 27th, 2018
Image of a wrist x-ray.

Enlarge (credit: Bo Mertz)

Treating patients effectively involves a combination of training and experience. That's one of the reasons that people have been excited about the prospects of using AI in medicine: it's possible to train algorithms using the experience of thousands of doctors, giving them more information than any single human could accumulate.

This week has provided some indications that software may be on the verge of living up to that promise, as two papers describe excellent preliminary results with using AI for both diagnosis and treatment decisions. The papers involve very different problems and approaches, which suggests that the range of situations where AI could prove useful is very broad.

Choosing treatments

One of the two studies focuses on sepsis, which occurs when the immune system mounts an excessive response to an infection. Sepsis is apparently the third leading cause of death worldwide, and it remains a problem even when the patient is already hospitalized. There are guidelines available for treating sepsis patients, but the numbers suggest there's still considerable room for improvement. So a small UK-US team decided to see if software could help provide some of that improvement.

Read 9 remaining paragraphs | Comments

Posted in AI, Computer science, deep learning, diagnosis, medicine, science, sepsis, treatment | Comments (0)