Connect with us

FACEBOOK

Computer vision inches toward ‘common sense’ with Facebook’s latest research

Published

on

Machine learning is capable of doing all sorts of things as long as you have the data to teach it how. That’s not always easy, and researchers are always looking for a way to add a bit of “common sense” to AI so you don’t have to show it 500 pictures of a cat before it gets it. Facebook’s newest research takes a big step toward reducing the data bottleneck.

The company’s formidable AI research division has been working for years now on how to advance and scale things like advanced computer vision algorithms, and has made steady progress, generally shared with the rest of the research community. One interesting development Facebook has pursued in particular is what’s called “semi-supervised learning.”

Generally when you think of training an AI, you think of something like the aforementioned 500 pictures of cats — images that have been selected and labeled (which can mean outlining the cat, putting a box around the cat or just saying there’s a cat in there somewhere) so that the machine learning system can put together an algorithm to automate the process of cat recognition. Naturally if you want to do dogs or horses, you need 500 dog pictures, 500 horse pictures, etc. — it scales linearly, which is a word you never want to see in tech.

Semi-supervised learning, related to “unsupervised” learning, involves figuring out important parts of a data set without any labeled data at all. It doesn’t just go wild, there’s still structure; for instance, imagine you give the system a thousand sentences to study, then showed it 10 more that have several of the words missing. The system could probably do a decent job filling in the blanks just based on what it’s seen in the previous thousand. But that’s not so easy to do with images and video — they aren’t as straightforward or predictable.

But Facebook researchers have shown that while it may not be easy, it’s possible and in fact very effective. The DINO system (which stands rather unconvincingly for “DIstillation of knowledge with NO labels”) is capable of learning to find objects of interest in videos of people, animals and objects quite well without any labeled data whatsoever.

Animation showing four videos and the AI interpretation of the objects in them.

Image Credits: Facebook

It does this by considering the video not as a sequence of images to be analyzed one by one in order, but as a complex, interrelated set, like the difference between “a series of words” and “a sentence.” By attending to the middle and the end of the video as well as the beginning, the agent can get a sense of things like “an object with this general shape goes from left to right.” That information feeds into other knowledge, like when an object on the right overlaps with the first one, the system knows they’re not the same thing, just touching in those frames. And that knowledge in turn can be applied to other situations. In other words, it develops a basic sense of visual meaning, and does so with remarkably little training on new objects.

This results in a computer vision system that’s not only effective — it performs well compared with traditionally trained systems — but more relatable and explainable. For instance, while an AI that has been trained with 500 dog pictures and 500 cat pictures will recognize both, it won’t really have any idea that they’re similar in any way. But DINO — although it couldn’t be specific — gets that they’re similar visually to one another, more so anyway than they are to cars, and that metadata and context is visible in its memory. Dogs and cats are “closer” in its sort of digital cognitive space than dogs and mountains. You can see those concepts as little blobs here — see how those of a type stick together:

Animated diagram showing how concepts in the machine learning model stay close together.

Image Credits: Facebook

This has its own benefits, of a technical sort we won’t get into here. If you’re curious, there’s more detail in the papers linked in Facebook’s blog post.

There’s also an adjacent research project, a training method called PAWS, which further reduces the need for labeled data. PAWS combines some of the ideas of semi-supervised learning with the more traditional supervised method, essentially giving the training a boost by letting it learn from both the labeled and unlabeled data.

Facebook of course needs good and fast image analysis for its many user-facing (and secret) image-related products, but these general advances to the computer vision world will no doubt be welcomed by the developer community for other purposes.

TechCrunch

FACEBOOK

5 Android apps you shouldn’t miss this week

Published

on

Apex Legends Mobile Cinematic Scene 7

Joe Hindy / Android Authority

Welcome to the 470th edition of Android Apps Weekly. Here are the big headlines from the last week.

  • YouTube Music has an annoying censorship bug on Nest Hubs. It doesn’t let you play music with sensitive album art. You get the same warning on the phone app, but you can usually bypass it. Unfortunately, there are limited ways to bypass it on your Nest Hub. Hit the link to learn more.
  • A former Facebook employee says Facebook can intentionally kill your battery. It does so through a process termed negative testing, where the app acts out, tanks your battery, and Facebook collects the data with it. It doesn’t happen to a ton of people, but it can happen to anyone.
  • Samsung updated Good Lock this week, just in time for its Samsung Galaxy S23 launch. The update added an option to update every installed plugin at once. Previously, you had to update each one individually. It’s a minor quality-of-life improvement, but it’s a welcome one.
  • ChatGPT is getting more serious. You can now spend $20 per month for a more powerful version of OpenAI’s bot. It’s only available to US customers right now, but it may expand later. The bot is also causing waves at Google, causing the company to ramp up its own AI work.
  • Apex Legends Mobile is shutting down after less than one year. EA made the announcement just a month after half of the Internet, including us, dubbed it the best new game of 2022. EA cites challenges with the content pipeline. It makes sense, since many of the newer updates have included a host of bugs that the developers just can’t seem to squash. Oh well, it was a nice run.

Pompom: The Great Space Rescue

Price: Free / $5.49

Pompom: The Great Space Rescue is a platformer. You play as Pompom and you progress through the game by jumping through and around obstacles, avoiding enemies, and solve puzzles to progress. It pays ode to the 16-bit era of gaming, so you’ll see a lot of elements, including graphics, from that era. There are also a bunch of weapons and tools you’ll get to help you on your way. The actual gameplay has some runner elements where you run forward automatically, and that’s not a 16-bit era style, but the game is still fun.

Memori Note

Price: Free / $2.49

Memori Note screenshot 2023

Memori Note is a note-taking app with an emphasis on reminding you of things. You write down what you want in the app, ask it to remind you about it at a random time, and it’ll do just that. The app also has color coding, a tags and filters system, and we think it looks pretty nice with its muted colors. There are also some backup settings if you want to transfer notes to a new device. We’re not sure how well it’ll do long term, but it definitely has the potential.

Devil Hunter Idle

Price: Free to play

Devil Hunter Idle is an action idle game. Your character hacks and slashes its way to level-ups, loot, and resources. You use those resources to strengthen your character so they can go back out and hack and slash more bad guys. That’s the primary gameplay loop, and it plays similarly to classic games like Buff Knight. The game’s over-the-top art style makes it feel like a lot more is happening, and the player does get to control some aspects of combat. The advertising is annoying, but you can pay to remove all of them. Other than that and some early bugs, the game is decent for its genre.

Rewind: Music Time Travel

Price: Free

Rewind Music Time Travel screenshot 2023

Rewind: Music Time Travel is an app for music rediscovery. It’s basically a big timeline that you scroll through to see what the music world looked like in any given year. It’s a neat way to rediscover old hits, and remind yourself of stuff you used to listen to. When I tested this one, I used it to help fill out my YouTube Music library a little bit since I had forgotten some of the songs I used to listen to. This isn’t something you’ll use long-term, but it’s a neat little app anyway.

Checkers Clash

Price: Free to play

Checkers Clash is an online competitive game where you play checkers. It’s not a complicated experience. You get into a game with an opponent. The two of you take turns until one of you runs out of pieces or concedes the match. You can also invite your friends and play against them as well. Some other game features include 8×8 and 10×10 board options, bots to play against to improve your skill, and a rewards system where you collect various things. The matchmaking system is imperfect, as it is in almost all online games, but it’s one of the few competitive checkers apps on mobile.


If we missed any big Android apps or games releases, tell us about it in the comments.
Thank you for reading. Try these out too:

Source link

Continue Reading

FACEBOOK

Family and friends question police theory that Nicola Bulley fell in river

Published

on

Family and friends question police theory that Nicola Bulley fell in river
N

icola Bulley’s family and friends have questioned a police theory that the missing dog walker fell in a river.

In a Facebook post, Ms Bulley’s sister, Louise Cunningham, urged people to “keep an open mind” as there is “no evidence whatsoever” that the dog walker fell in the river.

“Off the back of the latest Police media update, please can I add there is no evidence whatsoever that she has gone into the river, it’s just a theory,” she said.

“Everyone needs to keep an open mind as not all CCTV and leads have been investigated fully, the police confirmed the case is far from over.”

Search teams from Lancashire Constabulary are continuing to trawl the River Wyre near St Michael’s, working on the hypothesis that the missing mother-of-two, from nearby Inskip, could have fallen in when she disappeared on January 27.

Ms Bulley’s friend, Emma White, also cast doubt on the police theory, telling Sky News it was based on “limited information”.

“When we are talking about a life we can’t base it on a hypothesis – surely we need this factual evidence,” she said.

“That’s what the family and all of us are holding on to – that we are sadly no further on than last Friday.

“We still have no evidence, and that’s why we’re out together in force.

“You don’t base life on a hypothesis.”

Police believe the 45-year-old mortgage adviser went missing in just “a 10-minute window” while she was walking her dog, Willow, close to the River Wyre, after dropping off her daughters – aged six and nine – at school.

Ms Bulley had logged in to a Microsoft Teams call at 9.01am, which ended at 9.30am with her phone still connected to the call.

She was seen by another dog walker at 9.10am – the last known sighting – and police traced telephony records of her mobile phone as it remained on a bench overlooking the river at 9.20am.

The device was found by a dog walker at around 9.35am, with Willow nearby.

The police search has been aided by specialists and divers from HM Coastguard, mountain rescue, and Lancashire Fire and Rescue Service – with sniffer dogs, drones and police helicopters deployed.

Detectives are also working behind the scenes to analyse CCTV and dashcam videos, and members of the public with footage which could be useful have been urged to come forward.

Source link

Continue Reading

FACEBOOK

The Strange Reason I Facebook-Stalk My Ex’s New Girlfriend

Published

on

YourTango

I think we can all agree that the internet has made it much, much harder to get over a breakup.

Sure, you may have successfully erased his number from your phone, used his junior high football T-shirt as a rag, put away all your couple photos, ordered him never to call again, and cursed him to hell, but all of that effort is almost a waste considering he’s just a click of the mouse away. 

After a while, though, watching how he’s growing in the midriff via Facebook photos loses its luster. You already know almost everything about him anyway, after all that time you spent/wasted. But what about his new girlfriend? She’s someone to be curious about. 

RELATED: I Hired A Private Investigator To Spy On My Ex’s New Girlfriend

I’m not proud of this—as I am not proud of oh-so-many things I’ve admitted to but I am kind of obsessed with my ex’s new girlfriend.

Not in an Alicia Silverstone in The Crush sort of way—I am not going to trap the chick in a shed and set a hive of bees on her to attack. That would be crazy!

No, I just like to look at her Facebook page sometimes. It’s been set to private ever since the day I told my ex that I suspected he was lying to me about his relationship with her. He must have told her I was onto them because suddenly I didn’t have as much access to this mysterious girl I’d only ever met twice, who slipped in and changed my life without me noticing. 

I don’t blame her entirely, or even mostly. I blame him and, to a certain degree, I blame myself. But, I mean, I kind of blame her. We met! Twice! My ex made a point of telling me I would like her! (Red flag, FYI, ladies!) She seemed enthused to meet me! She seemed nice.

Two weeks before my ex pulled the plug on our relationship—and coincidentally the day before the two of them went on a business trip together—she even had drinks with us and his mom. It never occurred to me that something was amiss. 

Source link

Continue Reading

Trending

en_USEnglish