[[ACTUALITE] Control
Virginie Fayad
Images for download on the MIT News office website are made available to non-commercial entities, press and the general public under a Creative Commons Attribution Non-Commercial No Derivatives license. You may not alter the images provided, other than to crop them to size. A credit line must be used when reproducing images; if one is not provided below, credit the images to "MIT."
Researchers from MIT and Stanford University have devised a new machine-learning approach that could be used to control a robot, such as a drone or autonomous vehicle, more effectively and efficiently in dynamic environments where conditions can change rapidly.
This technique could help an autonomous vehicle learn to compensate for slippery road conditions to avoid going into a skid, allow a robotic free-flyer to tow different objects in space, or enable a drone to closely follow a downhill skier despite being buffeted by strong winds.
Additional authors of the paper are Jean-Jacques Slotine, professor of mechanical engineering and of brain and cognitive sciences at MIT, and Marco Pavone, associate professor of aeronautics and astronautics at Stanford. The research will be presented at the International Conference on Machine Learning (ICML).
A controller is the logic that enables a drone to follow a desired trajectory, for example. This controller would tell the drone how to adjust its rotor forces to compensate for the effect of winds that can knock it off a stable path to reach its goal.
Modeling a system by hand intrinsically captures a certain structure based on the physics of the system. For instance, if a robot were modeled manually using differential equations, these would capture the relationship between velocity, acceleration, and force. Acceleration is the rate of change in velocity over time, which is determined by the mass of and forces applied to the robot.
The team from MIT and Stanford developed a technique that uses machine learning to learn the dynamics model, but in such a way that the model has some prescribed structure that is useful for controlling the system.
When they tested this approach, their controller closely followed desired trajectories, outpacing all the baseline methods. The controller extracted from their learned model nearly matched the performance of a ground-truth controller, which is built using the exact dynamics of the system.
The researchers also found that their method was data-efficient, which means it achieved high performance even with few data. For instance, it could effectively model a highly dynamic rotor-driven vehicle using only 100 data points. Methods that used multiple learned components saw their performance drop much faster with smaller datasets.
In the future, the researchers are interested in developing models that are more physically interpretable, and that would be able to identify very specific information about a dynamical system, Richards says. This could lead to better-performing controllers.
As heated debate continues over how social media sites can improve the quality of news on their platforms while enforcing rules fairly, most Americans are pessimistic about these efforts and are highly concerned about several issues when it comes to social media and news.
Majorities say that social media companies have too much control over the news on their sites, and that the role social media companies play in delivering the news on their sites results in a worse mix of news for users. At the same time, social media is now a part of the news diet of an increasingly large share of the U.S. population.
The largest social media platforms control the content on their feeds using computer algorithms that rank and prioritize posts and other content tailored to the interests of each user. These sites allow users to customize these settings, though previous research has found that many Americans feel uncertain about why certain posts appear in their news feed on Facebook specifically. Social media companies have also been public about their efforts to fight both false information and fake accounts on their sites.
While social media companies say these efforts are meant to make the news experience on their sites better for everyone, most Americans think they just make things worse. A majority (55%) say that the role social media companies play in delivering the news on their sites results in a worse mix of news. Only a small share (15%) say it results in a better mix of news, while about three-in-ten (28%) think their efforts make no real difference.
While most Americans are pessimistic about the control social media companies have over the news people see, Republicans tend to be more negative than Democrats. Three-quarters of Republicans and Republican-leaning independents say social media companies have too much control over the mix of news that people see, compared with about half (53%) of Democrats and Democratic leaners. More Republicans (66%) than Democrats (49%) also say that these efforts result in a worse mix of news for users.
A large majority of Americans believe that social media companies favor some news organizations over others. About eight-in-ten U.S. adults (82%) say social media sites treat some news organizations differently than others, about five times the share saying all news organizations are treated the same (16%).
Social media companies do have established policies when it comes to publishers, including prioritizing certain news sources, banning or limiting others that produce lower-quality content, and using their monetization policies to discourage particular behaviors.
Among those U.S. adults who say social media companies treat some news organizations differently than others, there is broad agreement that they favor three types: those that produce attention-grabbing articles (88%), those with a high number of social media followers (84%) and those whose coverage has a certain political stance (79%).
While large social media companies have announced initiatives to favor high-quality news publishers in an effort to improve the news on their sites, fewer who say some news organizations are treated differently believe social media companies favor organizations that are well-established (56%), have high reporting standards (34%) or have politically neutral coverage (18%).
But in spite of the public discussion around potential censorship and efforts to monitor it, Americans are more concerned with the overall low quality of news available on social media sites. Of the seven issues asked about, about half of U.S. adults say that one-sided news (53%) and inaccurate news (51%) are very big problems when it comes to news on social media. Fewer say that censorship of the news (35%) or news organizations or personalities being banned (24%) are very big problems.
As large majorities say that the tone of American political debate has become more negative in recent years, about a third of U.S. adults (35%) say that uncivil discussions about the news are a very big problem when it comes to news on social media. Additionally, about a quarter (27%) say that the harassment of journalists is a very big problem associated with news on social media.
Republicans and Democrats disagree somewhat about which issues on social media are very big problems, especially when it comes to censorship and harassment. Republicans and Republican leaners are more likely to see censorship of the news as a very big problem on social media (43%) than Democrats and Democratic leaners (30%). Democrats, on the other hand, are about twice as likely as Republicans to say that harassment of journalists is a very big problem (36% vs. 17%). Despite these differences, one-sided news and inaccurate news top the list among both Republicans and Democrats.
Republicans are more likely to see a liberal lean than Democrats, a plurality of whom describe the news on social media as moderate. Among social media news consumers, Republicans are more likely to say that the news they see leans liberal or very liberal (64%) than are Democrats (37%). And social media news consumers who are conservative Republicans are especially likely to say that the news leans left: 73% describe the news posts they see as liberal or very liberal.
I have an Orbi RBR850. Everything working well, have >100 devices with IP reservations, hence I have only a small DHCP address range set aside for "other devices". Recently I've been seen an apple device (with a private MAC address, always the same though) attaching to my network. ALL of my devices are accounted for - it's not one I own. So, I blocked that MAC address in Access Control.
However, I still see it attaching to my network and getting an IP address. If i change the address range of the DHCP, that rogue device still shows as attached with a new address in the new range. I want to block it from attaching to my network, period. Yes, I know I can change the SSID or SSID password but with the number of IoT devices I have, I don't prefer that solution as it will be a huge pain in the arse.
So, forum, what does access control actually *do*? Just block internet access? If so, how does one put a MAC address on a disallow list so it cannot attach to the network. Again, I know I can change SSID password or name - that's not an easy change given the number of devices I'd have to deal with
The log files weren't terribly self explanatory - the IP address in question shows up in the files, but not sure where exactly to look for actual packet traffic - anything you can suggest would be appreciated.
On the access control, yes I have blocked the MAC - but it's still getting onto the Wifi so want to see what it's talking to, if anything. I'd prefer if Access Control MAC filtering disallowed access to Wifi entirely.
Thanks to all for the help - the rogue device is still showing as connected, but packet trace shows no traffic to or from other than the orbi doing name queries. I'm going to keep an eye on it for now. It's annoying though - block ought to mean "don't allow connectivity"
795a8134c1