Connect with us

Uncategorized

Deep Science: Alzheimer’s screening, forest-mapping drones, machine learning in space, more

Published

on

Research papers come out far too rapidly for anyone to read them all, especially in the field of machine learning, which now affects (and produces papers in) practically every industry and company. This column aims to collect the most relevant recent discoveries and papers — particularly in but not limited to artificial intelligence — and explain why they matter.

This week, a startup that’s using UAV drones for mapping forests, a look at how machine learning can map social media networks and predict Alzheimer’s, improving computer vision for space-based sensors and other news regarding recent technological advances.

Predicting Alzheimer’s through speech patterns

Machine learning tools are being used to aid diagnosis in many ways, since they’re sensitive to patterns that humans find difficult to detect. IBM researchers have potentially found such patterns in speech that are predictive of the speaker developing Alzheimer’s disease.

The system only needs a couple minutes of ordinary speech in a clinical setting. The team used a large set of data (the Framingham Heart Study) going back to 1948, allowing patterns of speech to be identified in people who would later develop Alzheimer’s. The accuracy rate is about 71% or 0.74 area under the curve for those of you more statistically informed. That’s far from a sure thing, but current basic tests are barely better than a coin flip in predicting the disease this far ahead of time.

This is very important because the earlier Alzheimer’s can be detected, the better it can be managed. There’s no cure, but there are promising treatments and practices that can delay or mitigate the worst symptoms. A non-invasive, quick test of well people like this one could be a powerful new screening tool and is also, of course, an excellent demonstration of the usefulness of this field of tech.

(Don’t read the paper expecting to find exact symptoms or anything like that — the array of speech features aren’t really the kind of thing you can look out for in everyday life.)

So-cell networks

Making sure your deep learning network generalizes to data outside its training environment is a key part of any serious ML research. But few attempt to set a model loose on data that’s completely foreign to it. Perhaps they should!

Researchers from Uppsala University in Sweden took a model used to identify groups and connections in social media, and applied it (not unmodified, of course) to tissue scans. The tissue had been treated so that the resultant images produced thousands of tiny dots representing mRNA.

Normally the different groups of cells, representing types and areas of tissue, would need to be manually identified and labeled. But the graph neural network, created to identify social groups based on similarities like common interests in a virtual space, proved it could perform a similar task on cells. (See the image at top.)

“We’re using the latest AI methods — specifically, graph neural networks, developed to analyze social networks — and adapting them to understand biological patterns and successive variation in tissue samples. The cells are comparable to social groupings that can be defined according to the activities they share in their social networks,” said Uppsala’s Carolina Wählby.

It’s an interesting illustration not just of the flexibility of neural networks, but of how structures and architectures repeat at all scales and in all contexts. As without, so within, if you will.

Drones in nature

The vast forests of our national parks and timber farms have countless trees, but you can’t put “countless” on the paperwork. Someone has to make an actual estimate of how well various regions are growing, the density and types of trees, the range of disease or wildfire, and so on. This process is only partly automated, as aerial photography and scans only reveal so much, while on-the-ground observation is detailed but extremely slow and limited.

Treeswift aims to take a middle path by equipping drones with the sensors they need to both navigate and accurately measure the forest. By flying through much faster than a walking person, they can count trees, watch for problems and generally collect a ton of useful data. The company is still very early-stage, having spun out of the University of Pennsylvania and acquired an SBIR grant from the NSF.

“Companies are looking more and more to forest resources to combat climate change but you don’t have a supply of people who are growing to meet that need,” Steven Chen, co-founder and CEO of Treeswift and a doctoral student in Computer and Information Science (CIS) at Penn Engineering said in a Penn news story. “I want to help make each forester do what they do with greater efficiency. These robots will not replace human jobs. Instead, they’re providing new tools to the people who have the insight and the passion to manage our forests.”

Another area where drones are making lots of interesting moves is underwater. Oceangoing autonomous submersibles are helping map the sea floor, track ice shelves and follow whales. But they all have a bit of an Achilles’ heel in that they need to periodically be picked up, charged and their data retrieved.

Purdue engineering professor Nina Mahmoudian has created a docking system by which submersibles can easily and automatically connect for power and data exchange.

A yellow marine robot (left, underwater) finds its way to a mobile docking station to recharge and upload data before continuing a task. (Purdue University photo/Jared Pike)

The craft needs a special nosecone, which can find and plug into a station that establishes a safe connection. The station can be an autonomous watercraft itself, or a permanent feature somewhere — what matters is that the smaller craft can make a pit stop to recharge and debrief before moving on. If it’s lost (a real danger at sea), its data won’t be lost with it.

You can see the setup in action below:

https://youtu.be/kS0-qc_r0

Sound in theory

Drones may soon become fixtures of city life as well, though we’re probably some ways from the automated private helicopters some seem to think are just around the corner. But living under a drone highway means constant noise — so people are always looking for ways to reduce turbulence and resultant sound from wings and propellers.

Computer model of a plane with simulated turbulence around it.

It looks like it’s on fire, but that’s turbulence.

Researchers at the King Abdullah University of Science and Technology found a new, more efficient way to simulate the airflow in these situations; fluid dynamics is essentially as complex as you make it, so the trick is to apply your computing power to the right parts of the problem. They were able to render only flow near the surface of the theoretical aircraft in high resolution, finding past a certain distance there was little point knowing exactly what was happening. Improvements to models of reality don’t always need to be better in every way — after all, the results are what matter.

Machine learning in space

Computer vision algorithms have come a long way, and as their efficiency improves they are beginning to be deployed at the edge rather than at data centers. In fact it’s become fairly common for camera-bearing objects like phones and IoT devices to do some local ML work on the image. But in space it’s another story.

Image Credits: Cosine

Performing ML work in space was until fairly recently simply too expensive power-wise to even consider. That’s power that could be used to capture another image, transmit the data to the surface, etc. HyperScout 2 is exploring the possibility of ML work in space, and its satellite has begun applying computer vision techniques immediately to the images it collects before sending them down. (“Here’s a cloud — here’s Portugal — here’s a volcano…”)

For now there’s little practical benefit, but object detection can be combined with other functions easily to create new use cases, from saving power when no objects of interest are present, to passing metadata to other tools that may work better if informed.

In with the old, out with the new

Machine learning models are great at making educated guesses, and in disciplines where there’s a large backlog of unsorted or poorly documented data, it can be very useful to let an AI make a first pass so that graduate students can use their time more productively. The Library of Congress is doing it with old newspapers, and now Carnegie Mellon University’s libraries are getting into the spirit.

CMU’s million-item photo archive is in the process of being digitized, but to make it useful to historians and curious browsers it needs to be organized and tagged — so computer vision algorithms are being put to work grouping similar images, identifying objects and locations, and doing other valuable basic cataloguing tasks.

“Even a partly successful project would greatly improve the collection metadata, and could provide a possible solution for metadata generation if the archives were ever funded to digitize the entire collection,” said CMU’s Matt Lincoln.

A very different project, yet one that seems somehow connected, is this work by a student at the Escola Politécnica da Universidade de Pernambuco in Brazil, who had the bright idea to try sprucing up some old maps with machine learning.

The tool they used takes old line-drawing maps and attempts to create a sort of satellite image based on them using a Generative Adversarial Network; GANs essentially attempt to trick themselves into creating content they can’t tell apart from the real thing.

Image Credits: Escola Politécnica da Universidade de Pernambuco

Well, the results aren’t what you might call completely convincing, but it’s still promising. Such maps are rarely accurate but that doesn’t mean they’re completely abstract — recreating them in the context of modern mapping techniques is a fun idea that might help these locations seem less distant.

Lyron Foster is a Hawaii based African American Musician, Author, Actor, Blogger, Filmmaker, Philanthropist and Multinational Serial Tech Entrepreneur.

Continue Reading
Comments

Uncategorized

A leading AI ethics researcher says she’s been fired from Google

Published

on

On December 2, the AI research community was shocked to learn that Timnit Gebru had been fired from her post at Google. Gebru, one of the leading voices in responsible AI research, is known among other things for her groundbreaking work in revealing the discriminatory nature of facial recognition, cofounding the Black in AI affinity group, and relentlessly advocating for diversity in the tech industry.

But on Wednesday evening, she announced on Twitter that she had been terminated from her position as Google’s ethical AI co-lead. “Apparently my manager’s manager sent an email [to] my direct reports saying she accepted my resignation. I hadn’t resigned,” she said.

In an interview with Bloomberg on Thursday, Gebru said that the firing happened after a protracted fight with her superiors over the publication of an AI ethics research paper. One of Gebru’s tweets and a later internal email from Jeff Dean, head of Google AI, suggest that the paper was critical of the environmental costs and embedded biases of large language models.

Gebru, who had written the paper with four Google colleagues and two external collaborators, had submitted it to a research conference being held next year. After an internal review, she was asked to retract the paper or remove the names of the Google employees. She responded that she would do so if her superiors met a series of conditions. If they could not, she would “work on a last date,” she said.

She also sent a frustrated email to an internal listserv, Google Brain Women and Allies, detailing the repeated hardships she’d experienced as a Black female researcher. “We just had a Black research all hands with such an emotional show of exasperation,” she wrote. “Do you know what happened since? Silencing in the most fundamental way possible.”

Gebru then went on a vacation and received a termination email from Megan Kacholia, the VP of engineering at Google Research, before her return. “Thanks for making your conditions clear,” the email stated, as tweeted by Gebru. “We cannot agree to #1 and #2 as you are requesting. We respect your decision to leave Google as a result, and we are accepting your resignation.” Her email to the listserv was “inconsistent with the expectations of a Google manager,” it continued. “As a result, we are accepting your resignation immediately, effective today.”

On Thursday morning, after an outpouring of support for Gebru on social media, Dean sent an internal email to Google’s AI group with his account of the situation. He said that Gebru’s paper “didn’t meet our bar for publication” because “it ignored too much relevant research.” He also said that Gebru’s conditions included “revealing the identities of every person who Megan and I had spoken to and consulted as part of the review of the paper and the exact feedback.”

“Given Timnit’s role as a respected researcher and a manager in our Ethical AI team, I feel badly that Timnit has gotten to a place where she feels this way about the work we’re doing,” he wrote. “I know we all genuinely share Timnit’s passion to make AI more equitable and inclusive.”

Neither Gebru, Dean, nor Google communications responded to requests for comment, and many details surrounding the exact progression of events, or cause of termination, remain unclear. As they continue to emerge, many have brought renewed attention to a November 30 tweet that Gebru pinned to the top of her profile. “Is there anyone working on regulation protecting Ethical AI researchers, similar to whistleblower protection?” it reads. “Because with the amount of censorship & intimidation that goes on towards people in specific groups, how does anyone trust any real research in this area can take place?”

Continue Reading

Uncategorized

Gift Guide: Games on every platform to get you through the long, COVID winter

Published

on

Welcome to TechCrunch’s 2020 Holiday Gift Guide! Need help with gift ideas? We’re here to help! We’ll be rolling out gift guides from now through the end of December. You can find our other guides right here.

It’s a great time to be a gamer — I mean, what else is there to do? And with the prospect of a long winter and lonely holiday season ahead of us, here’s a list of games on all the major platforms that you can really sink your teeth — and a few dozen hours — into.

Buying for a gamer and have no idea what’s worthwhile? Once you’ve figured out which gaming system is their platform of choice, any of these should be guaranteed wins.

This article contains links to affiliate partners where available. When you buy through these links, TechCrunch may earn an affiliate commission.

All major platforms:

Assassin’s Creed: Valhalla

Screenshot of Assassin's creed valhalla showing a viking on a british landscape

Image Credits: Ubisoft

I genuinely enjoyed AC: Odyssey’s gorgeous landscapes and main characters, but the game systems felt disconnected and arbitrary. That’s much less the case with AC: Valhalla, which tells a similarly sprawling tale of vikings in England but works a little harder to put it together into a cohesive whole. It’s still very much “Ubisoft Game, but with Vikings” but that’s not necessarily a bad thing.

Price: $50 from Amazon

Minecraft Dungeons

I thought this game was a bit limited when it first came out, but since then it has gotten several new areas and cross-platform multiplayer. Between that and its simplified systems and PG-level violence, Minecraft: Dungeons is a great option for families that want to fight monsters together.

Price: $20-30 (depending on platform) from Mojang

Call of Duty: Black Ops Cold War

Reviewers agree: the new CODBLOPS is definitely more CODBLOPS. The latest in the gritty military series is the one everyone will be playing for the next year, so it’s definitely a must-have for quite a few people.

Price: $50-60 (depending on platform) from Amazon

Cyberpunk 2077

Image Credits: CD Projekt Red

Cyberpunk 2077, the futuristic RPG from the creators of the Witcher, isn’t out yet, but it’s one of the most anticipated titles in recent years and your special someone might like the idea that they’re getting it day one. Of course if it’s anything like The Witcher 3, they’re probably going to want to wait a few months for the bugs to get ironed out. But hey, it’s an option.

Price: $50 from Amazon

PS4 and PS5

Spider-Man: Miles Morales

Image Credits: Sony

This semi-sequel to the much-lauded 2018 Spider-Man is smaller in scale but plays even better. Plus it has a wonderfully inclusive cast and tone and feels authentic where the original, for all its strengths, had a pretty flat take on New York. Bonus: If you buy the PS5 version of Miles Morales, you get the remastered 2018 game for free. I’d argue you’re simply not going to find a better bang for your buck right now with any other new game.

Price: $50 on Amazon

Demon’s Souls

Image Credits: Sony

The only “true” next-generation game out there right now is a remake of a PlayStation 3 game, and in many ways it feels like it. But in other ways, it’s the most amazing game on the market right now. If your loved one has enjoyed Dark Souls, Bloodborne, Sekiro, and other incredibly hard games, this is the one to get.

Price: $70 on Amazon

Ghost of Tsushima

Between Nioh 2, Sekiro, and Ghost of Tsushima, there’s a real bumper crop of samurai and ninja action to be had. But Ghost is the broadest and most beautiful of them all — if not necessarily the deepest.

What it lacks in challenge… first of all, is more than made up by the difficulty of those other two games I mentioned, give me a break. But Ghost’s draw is in the unity and beauty of its game world and systems. For example, instead of a quest marker or arrow pointing towards your objective, the wind is just always blowing in that direction. Amazing, right? The single player campaign is remarkably well acted, and a free update has brought a surprisingly extensive multiplayer co-op mode as well. This is truly a game you can lose yourself in. Just don’t start trying to collect everything or you’ll never leave the first area.

Price: $40 on Amazon

13 Sentinels: Aegis Rim

Image Credits: Vanillaware

This totally unique game came out of left field and obsessed me for two solid weeks. A combination of adventure game, visual novel, and tactical action game, 13 Sentinels puts you in charge of a bunch of high school kids piloting giant robots to save the world from alien invaders. (In case you can’t tell, it’s a Japanese production.)

Sound familiar? That’s the idea — and then it starts pulling rugs out from under you and doesn’t stop until the last few minutes. The labyrinthine story, which progresses simultaneously through 13 interwoven narratives, is the very best kind of sci-fi mind-boggler and a pleasure to unravel from start to finish. The combat is also compelling and satisfying, if not particularly deep or challenging. There’s simply nothing else like this out there.

Price: Currently $30 from GameStop

Xbox One and Series X

Halo: The Master Chief Collection

Image of Master Chief from halo

Image Credits: Microsoft

If your loved one is a Halo fan, they’re likely very sad since Halo: Infinite, once a launch title for the new console, won’t be coming out until next year. But it can’t hurt to have the original games all updated and beautified to play through as an appetizer. Plus there’s the famous Halo multiplayer to get everyone through the winter.

Price: $30 from Amazon

Yakuza: Like a Dragon

Image Credits: Sega

The latest in the long-running and beloved Yakuza series of character-driven adventures of Japanese gangsters set in a fictional Tokyo neighborhood, this one changes up the style with a turn-based combat system and new protagonist — but some are calling it the best yet.

Price: $35 from Amazon

Gears Tactics

Image Credits: Microsoft

No one really expected that the Gears of War series would lend itself to a tactics game in the style of XCOM — let alone that it would leapfrog others in the genre and become one of the best you can get, period. Naturally it isn’t quite the urgent, visceral experience that Gears normally is, but this is a surprisingly deep and engrossing game.

Price: $38 from Amazon

Ori and the Will of the Wisps

The sequel to the lauded “metroidvania” Ori and the Blind Forest is technically on several platforms, but the Series X seems to be the absolute best one on which to play it. With beautifully updated art and a silky-smooth framerate, this will look better on that new 4K HDR TV than many “real” next-generation games. But don’t let the beautiful yet cute art style make you think this is will be a cakewalk. Like the first in the series, you’ll need some serious dexterity to complete this platformer.

Price: $30 from Moon Studios

Risky move: Preorder Halo: Infinite

No one is quite sure whether the first Halo of the next generation is going to be as good as everyone hopes, and a delay to early next year didn’t allay anyone’s fears. That said, many a gamer will cherish the idea of playing the latest in this venerable series day one, so pre-ordering a copy is a possibility if none of the other games really ring their bell.

Price: $60 from Amazon

Switch

Mario Kart Live: Home Circuit

Technically this is also a “toy,” because it’s real-life RC carts zooming around your home on an augmented-reality racetrack. We thought it was tons of fun, and it’s a great way to take video games off the TV and into real life… kind of. Just be aware that every player needs their own cart and their own Switch.

Price: $99 from Best Buy

Hyrule Warriors: Age of Calamity

Don’t go into this game expecting a full-on new Zelda title, and you’ll do just fine. This is definitely an action game, and a big, rather mindless one at that. But it’s hard to resist the concept of playing as Link, Zelda, or any of the champions from Breath of the Wild and dispatching enemies by the hundreds.

Price: $50 on Amazon

Super Mario 3D All-Stars

Image Credits: Ninendo

Okay, I gave Nintendo some guff over the perfunctory nature of this collection of amazing games. I’ve wanted to replay Mario 64 for years and was waiting for Nintendo to touch it up just a bit — but it, and Super Mario Sunshine, and Super Mario Galaxy, are virtually unchanged in this retro package. Really, you couldn’t make it widescreen? But for most, the chance to play these games again (or for the first time) on the Switch is worth the price of admission, period.

Price: $60 from Amazon

PC

Spelunky 2 and/or Hades

The “roguelite” genre, with its randomly generated levels and complex interlocking systems, has grown in popularity and sophistication for years — and here we have two fine examples that take the genre in different directions.

Spelunky 2 is the most traditional, in a way. Sequel to one of the best games out there, this one adds more variety, more weirdness, and more challenge to the unforgiving platforming of the original. Like before, every death (and there will be a lot) is avoidable and while some runs may last only seconds, it’s hard to be deterred when you know that if you just paid a little more attention, or saved your bombs, or went over that other way… just one more game. (Pro tip: Buy a couple copies for friends and indulge in jolly cooperation.)

Hades combines the procedurally generated levels with an incredibly beautiful art style and an ingenious story and progression system. Escaping from the ever-shifting landscape of Hades, you’re going to die over and over, but as a young god that’s more inconvenience than obstacle. Meanwhile every death and every inch of progress moves you closer to the mystery of your birth in a clever modern take on Greek mythology. It’s honestly hard to imagine how Hades could be improved in almost any way.

Price: $20 for Spelunky 2 on Steam | $25 for Hades on Steam

Crusader Kings III

Image Credits: Paradox

This long-awaited strategy title puts you in the throne room of a European medieval dynasty, where you can do… pretty much anything to get ahead. Assassinations, proxy wars, brutal taxes, religious cannibalism, strategic marriages… it’s all on the table. This is a story-telling engine that’s remarkably robust and, once you get past the initial learning curve, very fun. It’s also very, very nerdy, and proud of it.

Price: $50 on Steam

Other options

Game & Watch: Super Mario Bros

Nintendo's Super Mario Bros handheld system

Image Credits: Devin Coldewey / TechCrunch

This little gadget has the original Super Mario Bros, its sequel (not the weird one — what we knew as “The Lost Levels”), and a remade LCD game all built in. It’s a charming device and the games play well, plus you can turn it off and resume progress later, making it that much easier to get through the whole game.

Price: $50 (but finding one in stock can be challenging.)

Backbone One for iPhone

Image Credits: Backbone

Got a friend who prefers to game on their phone? The Backbone is built for them. This snap-on controller brings buttons and analog triggers (and good ones, at that!) into the iOS gaming world, along with a surprisingly solid companion app that can do things like record your gameplay and help you edit and post your highlight reels. It only works with select iOS titles, but the library is growing. TechCrunch Editor-In-Chief Matthew Panzarino reviewed it in October and gave it his stamp of approval with very little reservation.

Price: $99 from Backbone

 

Continue Reading

Uncategorized

What about $30 billion under 30

Published

on

Hello and welcome back to Equity, TechCrunch’s venture capital-focused podcast (now on Twitter!), where we unpack the numbers behind the headlines.

We’re back with not an Equity Shot or Dive of Monday, this is just the regular show! So, we got back to our roots by looking at a huge number of early stage rounds. And a few other things that we were just too excited about to not mention.

So from Chris and Danny and Natasha and I, here’s the rundown:

That was a lot, but how could we leave any of it out? We’re back Monday with more!

Equity drops every Monday at 7:00 a.m. PDT and Thursday afternoon as fast as we can get it out, so subscribe to us on Apple PodcastsOvercastSpotify and all the casts.

Continue Reading

Trending