MIT researchers have used a new reinforcement learning system to teach robots how to acclimate to complex landscapes at high speeds, reports Emmett Smith for Mashable. “After hours of simulation training, MIT’s mini-cheetah robot broke a record with its fastest run yet,” writes Smith.
CSAIL researchers developed a new machine learning system to teach the MIT mini cheetah to run, reports James Vincent for The Verge. “Using reinforcement learning, they were able to achieve a new top-speed for the robot of 3.9m/s, or roughly 8.7mph,” writes Vincent.
Graduate student Matt Groh speaks with Scientific American reporter Sarah Vitak about his team’s work studying whether human detection or artificial intelligence is better at identifying deepfakes and misinformation online. “One of the things that we would suggest for the future development of these systems is trying to figure out ways to explain why the AI is making a decision,” says Groh.
Researchers from MIT and journalists from STAT conducted a months long investigation and found that “subtle shifts in data fed into popular health care algorithms — used to warn caregivers of impending medical crises — can cause their accuracy to plummet over time, raising the prospect AI could do more harm than good in many hospitals,” reports Casey Ross for STAT.
Popular Science reporter Tatyana Woodall writes that CSAIL researchers have developed electromagnetic bot blocks that can reconfigure into various shapes and could potentially help astronauts build in space. “The electromagnetic lining of the 3D printed frames allows cubes to seamlessly attract, repel, or even turn themselves off,” writes Wood. “One cube takes a little over an hour to make, and only costs 60 cents.”
Physics World reporter Tim Wogan writes that MIT researchers used machine learning techniques to identify a mysterious “X” particle in the quark–gluon plasma produced by the Large Hadron Collider. “Further studies of the particle could help explain how familiar hadrons such as protons and neutrons formed from the quark–gluon plasma believed to have been present in the early universe,” writes Wogan.
Using machine learning techniques, MIT researchers have detected “X particles” produced by the Large Hadron Collider, reports Rahul Rao for Popular Science. “The results tell us more about an artifact from the very earliest ticks of history, writes Rao. “Quark-gluon plasma filled the universe in the first millionths of a second of its life, before what we recognize as matter—molecules, atoms, or even protons or neutrons—had formed.”
Scientists have discovered “X-particles” in the aftermath of collisions produced in the Large Hadron Collider, which could shed light on the structure of these elusive particles, reports Becky Ferreira for Vice. “X particles can yield broader insights about the type of environment that existed in those searing and turbulent moments after the Big Bang,” writes Ferreira.
STAT reporters Katie Palmer and Casey Ross spotlight how Prof. Regina Barzilay has developed an AI tool called Mirai that can identify early signs of breast cancer from mammograms. “Mirai’s predictions were rolled into a screening tool called Tempo, which resulted in earlier detection compared to a standard annual screening,” writes Palmer and Ross.
In an article for The Wall Street Journal about next generation technologies that can create and quantify personal health data, Laura Cooper spotlights Prof. Dina Katabi’s work developing a noninvasive device that sits in a person’s home and can help track breathing, heart rate, movement, gait, time in bed and the length and quality of sleep. The device “could be used in the homes of seniors and others to help detect early signs of serious medical conditions, and as an alternative to wearables,” writes Cooper.
An international team of scholars, including MIT researchers, has published a new study exploring the history and use of letterlocking, reports William J. Broad for The New York Times. The researchers note that they hope their work prompts “novel kinds of archival research, and allows even very well-known artefacts to be examined anew.”
Wall Street Journal reporter Mark Hulbert writes that a new study by MIT researchers finds that most investors “can do much better than the one-size-fits-all approach to equity allocations that target-date funds offer for your retirement portfolio.”
Prof. Regina Barzilay speaks with Good Morning America about her work developing a new AI tool that could “revolutionize early breast cancer detection” by identifying patients at high risk of developing the disease. “If this technology is used in a uniform way,” says Barzilay, “we can identify early who are high-risk patients and intervene.”
Wired reporter Matt Simon spotlights CSAIL’s ‘Evolution Gym,’ a virtual environment where robot design is entirely computer generated. “There’s a potential to find new, unexpected robot designs, and it also has potential to get more high-performing robots overall,” says Prof. Wojciech Matusik. “If you start from very, very basic structures, how much intelligence can you really create?”
STAT reporter Katie Palmer writes that MIT researchers have developed a new machine learning model that can "flag treatments for sepsis patients that are likely to lead to a ‘medical dead-end,/ the point after which a patient will die no matter what care is provided.”