Category: TECHNOLOGY
Persephone puts poop to work to explore and heal your gut microbiome
What goes on in that gastrointestinal tract of yours? Well, we have a general idea, but evidence is mounting that the gut and the microbes that live there play an important role in a huge variety of health issues. Persephone is a biotech startup that — with the help of $15 million and a lot of poop — is building a library of the human microbiome and assembling a best-of list of helpful life forms that could do everything from easing digestion to fighting serious disease.
The democratization of once exclusive and expensive tools like rapid genetic sequencing has nurtured a new generation of biotech companies and therapeutic approaches. In this case it’s taking a closer look at how everyone’s microbiome — the often unique set of microorganisms that live in and on our body and perform various tasks for mutual benefit — differs and what those differences mean for our health.
Co-founder Stephanie Culler came from a background of genetic work at Genomatica, where they were working on producing chemicals normally sourced from petroleum by modifying bacteria to make it via fermentation.
“It required 5 years of genetic engineering, but it worked — it allowed us to build a blueprint of how to engineer a bacteria,” she said. “Now we’re doing something similar using the same tools. But we used to map a single microbe, and now we’re doing it with multiple microbes, building a precision map of the whole gut biome.”
There’s a lack of fundamental understanding in this field, she explained. Despite evidence that it’s involved in many processes, there’s historically been insufficient data to answer questions like how the microbiome affects disease progress, the effectiveness of therapeutics, the development of the immune system, even things like allergies. Part of why is the difficulty of collecting enough raw research material.
Persephone’s all-in-one “poop kit.”
The company has trained machine learning models on large datasets it has compiled itself through the laborious collection of stool samples from a large number of people, both healthy and suffering from various conditions.
“It’s not easy to give poop samples — there’s a stigma,” Culler explained. “So we focused on how to get it done easily. The initial funds we got through Y Combinator [one of our favorites in 2018] set us up to develop that infrastructure, to get large volumes of patient data.”
The microbes within the samples were isolated, sequenced, and catalogued, then that data was combined with lots of other health records — blood tests, behavior, medications, and so on. Machine learning is an efficient way of sorting through that kind of noisy dataset, and it identified both patterns worth investigating and what Culler called “superheroes” among the microbes.
For instance, a certain strain or functional type of bacterium may die out in the gut in the years preceding a diagnosis of colorectal cancer. Why? No one knows, but you don’t have to know for that kind of early marker to save lives. And what if its presence could be reinstated? That could …read more
HiddenLayer emerges from stealth to protect AI models from attacks
As AI-powered services like OpenAI’s GPT-3 grow in popularity, they become an increasingly attractive attack vector. Even shielded behind an API, hackers can attempt to reverse-engineer the models underpinning these services or use “adversarial” data to tamper with them. According to Gartner, 30% of all AI cyberattacks in 2022 will leverage these techniques along with data poisoning, which involves injecting bad data into the dataset used to train models to attack AI systems.
As in any industry, fighting security threats is a never-ending task. But Chris Sestito claims that his platform, HiddenLayer, can simplify it for AI-as-a-service vendors by automatically identifying malicious activity against models and responding to attacks.
HiddenLayer today emerged from stealth with $6 million in seed funding from Ten Eleven Ventures, Secure Octane and other investors. Sestito, the former director of threat research at Cylance and VP of engineering at Qualys, co-founded the company several months ago with Tanner Burns and Jim Ballard. Burns and Ballard also worked at Qualys and Cylance and spent time together at BlackBerry, where Ballard was a data curation team lead and Burns was a threat researcher.
“Virtually all enterprise organizations have made significant resource contributions to machine learning to give themselves an advantage — whether that value is in the form of product differentiation, revenue generation, cost savings or efficiencies,” Sestito told TechCrunch in an email interview. “Adversarial machine learning attacks are capable of causing all of the same damage we’ve seen in traditional cyber attacks including exposing customer data and destroying production systems. In fact, at HiddenLayer, we believe we’re not far off from seeing machine learning models ransomed back to their organizations.”
HiddenLayer claims that its technology can defend models from attacks without the need to access any raw data or a vendor’s algorithms. By analyzing model interactions — in other words, the data fed into the model (e.g., a picture of cats) and the predictions that the model outputs (e.g., the caption “cats”) — to spot patterns that could be malicious, HiddenLayer can work “non-invasively” and without prior knowledge of training data, Sestito said.
“Adversarial machine learning attacks are not loud like ransomware — you have to be looking for them to catch them in time,” Sestito said. “HiddenLayer has focused on a research-first approach that will allow us to publish our findings and train the world to be prepared.”
Mike Cook, an AI researcher who’s a part of the Knives and Paintbrushes collective, said that it’s unclear whether HiddenLayer is doing anything “truly groundbreaking or new.” (Cook is unaffiliated with HiddenLayer.) Still, he notes that there’s a benefit to what HiddenLayer appears to be doing: trying to package up knowledge about attacks on AI and make them more widely accessible.
“The AI boom is still booming, but a lot of that knowledge about how modern machine learning works and how best to use it is still locked away mostly to people who have specialist knowledge. Memorable examples for me include researchers managing to extract …read more
IBM hopes a new error mitigation technique will help it get to quantum advantage
It felt like for a long time, the quantum computing industry avoided talking about ‘quantum advantage’ or ‘quantum supremacy,’ the point where quantum computers can solve problems that would simply take too long to solve on classical computers. To some degree, that’s because the industry wanted to avoid the hype that comes with that, but IBM today brought back talk about quantum advantage again by detailing how it plans to use a novel error mitigation technique to chart a path toward running the increasingly large circuits it’ll take to reach this goal — at least for a certain set of algorithms.
It’s no secret that quantum computers hate nothing more than noise. Qubits are fickle things, after all, and the smallest change in temperature or vibration can make them decohere. There’s a reason the current era of quantum computing is associated with ‘noisy intermediate-scale quantum (NISQ) technology.
The engineers at IBM and every other quantum computing company are making slow but steady strides toward reducing that noise on the hardware and software level, with IBM’s 65-qubit systems from 2020 now showing twice the coherence time compared to when they first launched, for example. The coherence time of IBM’s transmon superconducting qubits is now over 1ms. That may not sound like much — and some other companies have shown
But IBM is also taking another approach but betting on new error mitigation techniques, dubbed probabilistic error cancellation and zero noise extrapolation. At a very basic level, you can almost think of this as the quantum equivalent of the active noise cancellation in your headphones. The system regularly checks the system for noise and then essentially inverts those noisy circuits to enable it to create virtually error-free results.
IBM has now shown that this isn’t just a theoretical possibility but actually works in its existing systems. One disadvantage here is that there is quite a bit of overhead when you constantly sample these noisy circuits and that overhead is exponential in the number of qubits and the circuit depths. But that’s a tradeoff worth making, argues Jerry Chow, the Director of Hardware Development for IBM Quantum.
“Error mitigation is about finding ways to deal with the physical errors in certain ways, by learning about the errors and also just running quantum circuits in such a way that allows us to cancel them,” explained Chow. “In some ways, error correction is like the ultimate error mitigation, but the point is that there are techniques that are more near term with a lot of the hardware that we’re building that already provide this avenue. The one that we’re really excited about is called probabilistic error cancellation. And that one really is a way of trading off runtime — trading off running more circuits in order to learn about the noise that might be inherent to the system that is impacting your calculations.”
The system essentially inserts additional gates into existing circuits to sample the noise inherent in the system. And while the overhead increases exponentially with the …read more
Teaching home robots to learn by watching people
Robotic learning has quickly become of automation’s most vibrant categories — and understandably so. Programming a robot has traditionally required a lot of technical know-how, but what if there was a simpler way for non-programmers/roboticists to teach these systems to do what we want?
Imitation and reinforcement learning are two of the most popular methods at the moment. The first involves taking control of the robot to teach it to perform a task, while the second entails training a system on millions of images.
A number of researchers are exploring an even more intuitive method that effectively trains a system by watching a human complete a task. A team at Carnegie Mellon University is demonstrating the in-the-Wild Human Imitating Robot Learning, or WHIRL, an algorithm that can train a system by watching a video.
In their demos, an off-the-shelf mobile robotic arm learns to complete 20+ household chores, including opening and closing drawers and appliances and taking out the trash.
“Imitation is a great way to learn,” Robotics Institute PhD student Shikhar Bahl said in a release. “Having robots actually learn from directly watching humans remains an unsolved problem in the field, but this work takes a significant step in enabling that ability.”
It’s easy to see how a feature like this might prove particularly handy in a home setting, where roboticists anticipate these systems will one day be deployed to help elderly homeowners and other people with impaired movements.
In the case of WHIRL, no special adds-ons are required. The robot simply attempts to execute a certain task until it’s successful, even if it takes several times to fully master it. As CMU notes, its own approach may not be identical to the humans — instead, the system is looking to find the best method to complete the task based on its own hardware limitations.
Right now the system is trained by watching videos, and the team is looking to expand things out to include clips from services like YouTube.
https://techcrunch.com/2022/07/18/teaching-home-robots-to-learn-by-watching-people/
Africa’s startup market is bucking the global slowdown
Parsing the latest data on the startup fundraising market in Q2, TechCrunch has explored the global perspective, taken a closer look at fintech, asked how much dry powder VCs have and brought the latest from unicorn land. But we are not yet done.
Later this week, we’re looking at the European venture market, which is faring a little better than many other major startup hubs. But Europe is not doing the best — that honor may belong to Africa. Data indicates that Africa is not only posting year-over-year gains in venture capital fundraising, it could also be on track for a record year.
The Exchange explores startups, markets and money.
Read it every morning on TechCrunch+ or get The Exchange newsletter every Saturday.
Regional discrepancies abound in a changing global market, but today, instead of merely tracking which areas are declining the fastest when compared to 2021, we have some more positive data to chew on.
https://techcrunch.com/2022/07/18/africas-startup-market-is-bucking-the-global-slowdown/
NHTSA investigates another Tesla crash involving death of motorcyclist
The National Highway Traffic Safety Administration is opening yet another special investigation into a Tesla vehicle crash, according to documents viewed by TechCrunch. This time it involves the crash of a 2021 Tesla Model Y that killed a motorcyclist in California earlier this month.
Reuters was the first to report the special investigation.
This is the 38th special investigation of a crash involving a Tesla vehicle since 2016. Of those crashes, 18 were fatal. The latest probe, like most of the others, seeks to ascertain whether or not Autopilot, Tesla’s advanced driver assistance system, was in use at the time of the crash.
Earlier this month, the NHTSA opened an investigation into one such fatal crash in which a pedestrian was killed and involved a 2018 Tesla Model 3 in California. The NHTSA also opened up special probes into another fatal Tesla crash, this one in Florida, which killed a 66-year-old Tesla driver and a 67-year-old passenger. In May, the agency began investigating a crash involving a 2022 Tesla Model S that killed three people.
The NHTSA declined to comment on the case, as it is still open. Local news reported that on July 7, a 48-year-old motorcyclist was killed after a collision on the Riverside Freeway in Riverside. He was riding in the HOV lane and was approached from behind by the Tesla. A crash then occurred that hurled the rider onto the freeway, according to My News LA.
The NHTSA usually opens more than 100 special crash investigations each year that probe emerging technologies.
Tesla cannot be reached for comment because it has disbanded its press office.
Crypto exchange Gemini executes second round of layoffs less than two months after axing 10% of staff
Just seven weeks after crypto exchange Gemini cut approximately 10% of its workforce due to “turbulent market conditions,” the startup has made a second round of layoffs, TechCrunch has learned, and there may be more on the way.
The company had not widely communicated the extent of Monday’s layoffs internally, leaving employees to speculate on the exact number of co-workers laid off in this most recent downsizing. A source close to the company noted that there was a reduction of 7%, or 68 members, in Gemini’s companywide Slack channel Monday morning.
Gemini has not yet responded to a request for comment.
The same source, who spoke with TechCrunch under the condition of public anonymity, said that the company was laying off staff due to what it described as “extreme cost cutting.”
Last week, an internal operating plan document was shared around the office and on an anonymous professional network Blind on July 14, but taken down shortly after, the source shared. The document highlighted a plan that would take the company to about 800 employees, which was around 15% fewer than the 950 employees at the time.
“It’s come to my attention that at least one team member thinks it’s a good idea to post a snippet of our technology operating plan on a third party website (Blind),” Cameron Winklevoss, co-founder of Gemini, wrote in a Slack message at the time. “Wow, super lame … if you are leaking company information, you are exhibiting a low level of consciousness and respect for your fellow team members who greatly benefit from the openness we are trying to create and foster here.”
TechCrunch viewed an image of the Slack message in question. Winklevoss also said that the message was a “friendly reminder that Karma is the blockchain of the universe — an immutable ledger that keeps track of positive and negative behavior.”
“We are going to the moon. We are going to need cosmic consciousness to get there. Earthly consciousness will not be enough,” Winklevoss added. “If you are exhibiting the behavior of a first-time human, time to level-up or respectively bow out, if for no other reason but to avoid an expensive bill in the future.”
In recent months, a wave of crypto-focused companies have laid off employees. Last week, NFT marketplace OpenSea reduced its staff by 20%. A number of other firms including Crypto.com, Coinbase and Bybit also cut staff amid crypto market volatility.
This is a developing story and will be updated if new information becomes available.
Slack is increasing prices and changing the way its free plan works
Slack, the chat platform that serves as an online watercooler for oh-so-many teams, is bumping up its monthly price and changing the way its free plan works.
The company announced the changes via blog post this afternoon.
Here’s what’s changing:
- If you pay for the “Pro” plan by the month, the price will increase from $8 per user per month to $8.75 per user per month.
- If you pay for the “Pro” plan by the year, the price will increase from $6.67 per user per month to $7.25 per user per month.
- If you’re on the free plan, they’re changing the way/duration messages are saved. Previously, free Slacks would show the last 10,000 messages and 5 GB worth of uploads. Moving forward it’ll be based on time rather than amount, with Slack showing the last 90 days of messages/uploads regardless of how much or how little is sent.
(Prices above are for U.S. users, but it’ll increase worldwide; the price change chart for other countries is available here.)
Slack notes that the pricing changes only impact “Pro” plan users, so teams on the Business+ plan (which currently costs $12.50 or $15 per user per month) or custom enterprise plans do not seem to be impacted.
Slack says the pricing change — which it says is the “first price increase since [Slack] first launched in 2014” — will go into effect as of September 1st, 2022. The company’s blog post also notes that you can lock in the existing price for an additional year by renewing before September.
Look out Putin — Ukrainian woman invents ‘solar for balconies’ to wean Europe off Russian gas
WeDoSolar claims it can reduce a household’s CO2 footprint by up to 600 kg and electricity bills by up to 25% per year.
Much has been made of how European countries are, on the one hand, supporting Ukraine in its fight against the heinous and illegal invasion by Russia but at the same time remaining heavily dependent on Russian energy sources.
Many countries in the EU — currently experiencing a heatwave — have scrambled to switch to alternatives, such as renewables, which are considered one of the key factors of in reaching energy independence from Russia.
So it is a delicious irony that a Ukrainian woman tech entrepreneur has come up with a way to allow ordinary Europeans to fight Russia, by easily installing solar panels on their numerous balconies.
WeDoSolar, launched in February 2022 (the month Russia invaded Ukraine) and says it has so far received “thousand of orders” for its “vertical solar power” panels, which are specifically designed to be mounted, with weatherproof straps, onto to balconies by non-tech-savvy users.
Image Credits: WeDoSolar
WeDoSolar claims it can reduce a household’s CO2 footprint by up to 600 kg and electricity bills by up to 25% per year.
The EU imported approximately 155 billion cubic metres (bcm) of natural gas from Russia in 2021. That amounts to around 40% of the EU’s total gas consumption, with dependency rising to 65% in Germany.
In theory, if every balcony in Europe had solar panels such as WeDoSolar’s, it would make getting off Russian oil and gas a great deal easier.
Designed by German engineers and coming insured as part of the offer, the lightweight panels weigh 1 kg each and plug into a standard power socket. The WeDoSolar Microinverter then pushes the power into the home grid, allowing the panels to power home appliances ahead of using the normal grid, since solar is always used ahead of normal grid power, claims the company.
Users can also track the energy generation of their solar panels and the amount of CO2 saved in an app, which converts the metrics into language such as “60 phone charges” or 600 kg of yearly CO2 reduction equaling a “1,000 m2 forest.”
Focusing on EU markets, the eight-panel set costs €1,299 EUR or comes as a free rental model for electric vehicle owners in exchange for CO2 certificates. These are then bundled and sold to oil companies, as oil companies are obliged to buy certificates on a yearly basis from EV owners, at least in Germany.
WeDoSolar also offers employee-benefit programs to large enterprises where the employers get the benefit of CO2 certificates for their companies while the employees can potentially lower electricity bills (assuming the sun is shining).
The startup was founded by Karolina Attspodina, a Ukrainian-born serial entrepreneur and Qian Qin, a hardware engineer, serial entrepreneur and a former managing director of BSH Digital Ventures GmbH. The startup has been backed by Angel backers, but no institutional investors, as yet.
WeDoSolar has competitors in the shape of Yuma, Priwatt and Plugin Energy, however, says the company, it plans to compete by being …read more
Max Q: Jaw-dropping Wondrous Spectacular Telescope
Hello and welcome back to Max Q. In this issue:
- The first images from James Webb Space Telescope are here
- Lunar Outpost eyes up moon markets
- News from ABL Space Systems, Skyrora and more
By the way…we’re right around the corner from TC Sessions: Robotics, a full-day digital gathering of some of the world’s leading founders, technologists, engineers, researchers and investors in robotics and AI! Register and learn more here.
OK, on to the news.
The first images from James Webb Space Telescope are here and they do not disappoint
James Webb Space Telescope (JWST) is one of the most exciting astronomical missions in years. With JWST, scientists are hoping to learn more about the universe, from star formation to black holes to exoplanets.
This week, NASA released the first five images from the optical instrument and they did not disappoint. The first image, revealed by President Joe Biden, is a sublime photo of the cluster SMACS 0723. As NASA Administrator Bill Nelson put it, “If you held a grain of sand on the tip of your finger at arm’s length, that is the part of the universe that you’re seeing, just one little speck of the universe.”
Check out this story on what the other four images tell us about the universe and this deep dive from Devin Coldewey into just how a spacecraft a million miles away manages to send dozens of gigabytes of data back to Earth.
Image Credits: NASA (opens in a new window)
Lunar Outpost eyes up moon markets
I caught up with Lunar Outpost CEO Justin Cyrus a few months after the company closed a $12 million seed round, led by Explorer 1 Fund with participation from Promus Ventures, Space Capital, Type 1 Ventures and Cathexis Ventures.
Lunar Outpost is developing rovers and other technologies for terrestrial and space applications, and it already has a handful of space missions lined up for 2023 and beyond. For the first mission, Lunar Outpost will send a 10-kilogram rover, called the Mobile Autonomous Prospecting Platform (MAPP), to the moon’s south pole on a mission in partnership with Nokia and Intuitive Machines in early 2023. The company is also sending a rover to explore the mysterious Reiner Gamma feature on the moon in a fully funded mission for NASA, also in partnership with Intuitive Machines. To top it off, it’s part of a team that includes Northrop Grumman, Michelin, AVL and Intuitive Machines that’s bidding on a contract to build a crewed lunar terrain vehicle for NASA.
“We saw our competitors starting to try to catch up and we felt like this funding can not only be utilized to further differentiate ourselves, and dig a little bit of a moat, if you will, but also accelerate our commercialization timeline for cislunar space,” Cyrus said.
https://techcrunch.com/2022/07/18/max-q-jaw-dropping-wondrous-spectacular-telescope/