Tag: Science

Posted on

This box sucks pure water out of dry desert air

For many of us, clean, drinkable water comes right out of the tap. But for billions it’s not that simple, and all over the world researchers are looking into ways to fix that. Today brings work from Berkeley, where a team is working on a water-harvesting apparatus that requires no power and can produce water even in the dry air of the desert. Hey, if a cactus can do it, why can’t we?

While there are numerous methods for collecting water from the air, many require power or parts that need to be replaced; what professor Omar Yaghi has developed needs neither.

The secret isn’t some clever solar concentrator or low-friction fan — it’s all about the materials. Yaghi is a chemist, and has created what’s called a metal-organic framework, or MOF, that’s eager both to absorb and release water.

It’s essentially a powder made of tiny crystals in which water molecules get caught as the temperature decreases. Then, when the temperature increases again, the water is released into the air again.

Yaghi demonstrated the process on a small scale last year, but now he and his team have published the results of a larger field test producing real-world amounts of water.

They put together a box about two feet per side with a layer of MOF on top that sits exposed to the air. Every night the temperature drops and the humidity rises, and water is trapped inside the MOF; in the morning, the sun’s heat drives the water from the powder, and it condenses on the box’s sides, kept cool by a sort of hat. The result of a night’s work: 3 ounces of water per pound of MOF used.

That’s not much more than a few sips, but improvements are already on the way. Currently the MOF uses zicronium, but an aluminum-based MOF, already being tested in the lab, will cost 99 percent less and produce twice as much water.

With the new powder and a handful of boxes, a person’s drinking needs are met without using any power or consumable material. Add a mechanism that harvests and stores the water and you’ve got yourself an off-grid potable water solution.

“There is nothing like this,” Yaghi explained in a Berkeley news release. “It operates at ambient temperature with ambient sunlight, and with no additional energy input you can collect water in the desert. The aluminum MOF is making this practical for water production, because it is cheap.”

He says there are already commercial products in development. More tests, with mechanical improvements and including the new MOF, are planned for the hottest months of the summer.

Posted on

Forget DeepFakes, Deep Video Portraits are way better (and worse)

The strange, creepy world of “deepfakes,” videos (often explicit) with the faces of the subjects replaced by those of celebrities, set off alarm bells just about everywhere early this year. And in case you thought that sort of thing had gone away because people found it unethical or unconvincing, the practice is back with the highly convincing “Deep Video Portraits,” which refines and improves the technique.

To be clear, I don’t want to conflate this interesting research with the loathsome practice of putting celebrity faces on adult film star bodies. They’re also totally different implementations of deep learning-based image manipulation. But this application of technology is clearly here to stay and it’s only going to get better — so we had best keep pace with it so we don’t get taken by surprise.

Deep Video Portraits is the title of a paper submitted for consideration this August at SIGGRAPH; it describes an improved technique for reproducing the motions, facial expressions, and speech movements of one person using the face of another. Here’s a mild example:

What’s special about this technique is how comprehensive it is. It uses a video of a target person, in this case President Obama, to get a handle on what constitutes the face, eyebrows, corners of the mouth, background, and so on, and how they move normally.

Then, by carefully tracking those same landmarks on a source video it can make the necessary distortions to the President’s face, using their own motions and expressions as sources for that visual information.

So not only does the body and face move like the source video, but every little nuance of expression is captured and reproduced using the target person’s own expressions! If you look closely, even the shadows behind the person (if present) are accurate.

The researchers verified the effectiveness of this by comparing video of a person actually saying something on video with what the deep learning network produced using that same video as a source. “Our results are nearly indistinguishable from the real video,” says one of the researchers. And it’s true.

So, while you could use this to make video of anyone who’s appeared on camera appear to say whatever you want them to say — in your voice, it should be mentioned — there are practical applications as well. The video shows how dubbing a voice for a movie or show could be improved by syncing the character’s expression properly with the voice actor.

There’s no way to make a person do something or make an expression that’s too far from what they do on camera, though. For instance, the system can’t synthesize a big grin if the person is looking sour the whole time (though it might try and fail hilariously). And naturally there are all kinds of little bugs and artifacts. So for now the hijinks are limited.

But as you can see from the comparison with previous attempts at doing this, the science is advancing at a rapid pace. The differences between last year’s models and this years are clearly noticeable, and 2019’s will be more advanced still. I told you all this would happen back when that viral video of the eagle picking up the kid was making the rounds.

“I’m aware of the ethical implications,” coauthor Justus Theis told The Register. “That is also a reason why we published our results. I think it is important that the people get to know the possibilities of manipulation techniques.”

If you’ve ever thought about starting a video forensics company, now might be the time. Perhaps a deep learning system to detect deep learning-based image manipulation is just the ticket.

The paper describing Deep Video Portraits, from researchers at Technicolor, Stanford, the University of Bath, the Max Planck Institute for Informatics, and the Technical University of Munich, is available for you to read here on Arxiv.

Posted on

Watch a hard-working robot improvise to climb drawers and cross gaps

A robot’s got to know its limitations. But that doesn’t mean it has to accept them. This one in particular uses tools to expand its capabilities, commandeering nearby items to construct ramps and bridges. It’s satisfying to watch but, of course, also a little worrying.

This research, from Cornell and the University of Pennsylvania, is essentially about making a robot take stock of its surroundings and recognize something it can use to accomplish a task that it knows it can’t do on its own. It’s actually more like a team of robots, since the parts can detach from one another and accomplish things on their own. But you didn’t come here to debate the multiplicity or unity of modular robotic systems! That’s for the folks at the IEEE International Conference on Robotics and Automation, where this paper was presented (and Spectrum got the first look).

SMORES-EP is the robot in play here, and the researchers have given it a specific breadth of knowledge. It knows how to navigate its environment, but also how to inspect it with its little mast-cam and from that inspection derive meaningful data like whether an object can be rolled over, or a gap can be crossed.

It also knows how to interact with certain objects, and what they do; for instance, it can use its built-in magnets to pull open a drawer, and it knows that a ramp can be used to roll up to an object of a given height or lower.

A high-level planning system directs the robots/robot-parts based on knowledge that isn’t critical for any single part to know. For example, given the instruction to find out what’s in a drawer, the planner understands that to accomplish that, the drawer needs to be open; for it to be open, a magnet-bot will have to attach to it from this or that angle, and so on. And if something else is necessary, for example a ramp, it will direct that to be placed as well.

The experiment shown in this video has the robot system demonstrating how this could work in a situation where the robot must accomplish a high-level task using this limited but surprisingly complex body of knowledge.

In the video, the robot is told to check the drawers for certain objects. In the first drawer, the target objects aren’t present, so it must inspect the next one up. But it’s too high — so it needs to get on top of the first drawer, which luckily for the robot is full of books and constitutes a ledge. The planner sees that a ramp block is nearby and orders it to be put in place, and then part of the robot detaches to climb up and open the drawer, while the other part maneuvers into place to check the contents. Target found!

In the next task, it must cross a gap between two desks. Fortunately, someone left the parts of a bridge just lying around. The robot puts the bridge together, places it in position after checking the scene, and sends its forward half rolling towards the goal.

These cases may seem rather staged, but this isn’t about the robot itself and its ability to tell what would make a good bridge. That comes later. The idea is to create systems that logically approach real-world situations based on real-world data and solve them using real-world objects. Being able to construct a bridge from scratch is nice, but unless you know what a bridge is for, when and how it should be applied, where it should be carried and how to get over it, and so on, it’s just a part in search of a whole.

Likewise, many a robot with a perfectly good drawer-pulling hand will have no idea that you need to open a drawer before you can tell what’s in it, or that maybe you should check other drawers if the first doesn’t have what you’re looking for!

Such basic problem-solving is something we take for granted, but nothing can be taken for granted when it comes to robot brains. Even in the experiment described above, the robot failed multiple times for multiple reasons while attempting to accomplish its goals. That’s okay — we all have a little room to improve.

Posted on

Teens dump Facebook for YouTube, Instagram and Snapchat

A Pew survey of teens and the ways they use technology finds that kids have largely ditched Facebook for the visually stimulating alternatives of Snapchat, YouTube, and Instagram. Nearly half said they’re online “almost constantly,” which will probably be used as a source of FUD, but really is just fine. Even teens, bless their honest little hearts, have doubts about whether social media is good or evil.

The survey is the first by Pew since 2015, and plenty has changed. The one that has driven the most change seems to be the ubiquity and power of smartphones, which 95 percent of respondents said they had access to. Fewer, especially among lower income families, had laptops and desktops.

This mobile-native cohort has opted for mobile-native content and apps, which means highly visual and easily browsable. That’s much more the style on the top three apps: YouTube takes first place with 85 percent reporting they use it, then Instagram at 72 percent, and Snapchat at 69.

Facebook, at 51 percent, is a far cry from the 71 percent who used it back in 2015, when it was top of the heap by far. Interestingly, the 51 percent average is not representative of any of the income groups polled; 36 percent of higher income households used it, while 70 percent of teens from lower income households did.

What could account for this divergence? The latest and greatest hardware isn’t required to run the top three apps, nor (necessarily) an expensive data plan. With no data to go on from the surveys and no teens nearby to ask, I’ll leave this to the professionals to look into. No doubt Facebook will be interested to learn this — though who am I kidding, it probably knows already. (There’s even a teen tutorial.)

Twice as many teens reported being “online constantly,” but really, it’s hard to say when any of us is truly “offline.” Teens aren’t literally looking at their phones all day, much as that may seem to be the case, but they — and the rest of us — are rarely more than a second or two away from checking messages, looking something up, and so on. I’m surprised the “constantly” number isn’t higher, honestly.

Gaming is still dominated by males, almost all of whom play in some fashion, but 83 percent of teen girls also said they gamed, so the gap is closing.

Phone-addicted teens aren’t as happy as those who play sports and hang out IRL, new study suggests

When asked whether social media had a positive or negative effect, teens were split. They valued it for connecting with friends and family, finding news and information, and meeting new people. But they decried its use in bullying and spreading rumors, its complicated effect on in-person relationships, and how it distracts from and distorts real life.

Here are some quotes from real teens demonstrating real insight.

Those who feel it has an overall positive effect:

  • “I feel that social media can make people my age feel less lonely or alone. It creates a space where you can interact with people.”
  • “My mom had to get a ride to the library to get what I have in my hand all the time. She reminds me of that a lot.”
  • “We can connect easier with people from different places and we are more likely to ask for help through social media which can save people.”
  • “It has given many kids my age an outlet to express their opinions and emotions, and connect with people who feel the same way.”

And those who feel it’s negative:

  • “People can say whatever they want with anonymity and I think that has a negative impact.”
  • “Gives people a bigger audience to speak and teach hate and belittle each other.”
  • “It makes it harder for people to socialize in real life, because they become accustomed to not interacting with people in person.”
  • “Because teens are killing people all because of the things they see on social media or because of the things that happened on social media.”

That last one is scary.

You can read the rest of the report and scrutinize Pew’s methodology here.

Posted on

Trump’s visa restrictions aimed at Chinese STEM students to start in June

In a policy change set for next month, the Trump administration is moving to shorten visas for Chinese students in fields like tech and engineering. While most visas are issued for the longest possible length of time under law, the new policy will allow U.S. officials to put a one-year cap on visas for Chinese graduate students who are “studying in fields like robotics, aviation and high-tech manufacturing,” according to the Associated Press.

A State Department official told The Hill that “Although the large majority of visas issued to Chinese nationals are issued for the maximum validity, consular officers may limit the validity of visas on a case-by-case basis” under the new rules.

Beyond the student limits, U.S. consulates and embassies reportedly received instructions that any Chinese citizen applying for a visa will need to secure additional special permission form the U.S. if they work in research or management for any company the U.S. Commerce Department lists as an entity “requiring higher scrutiny.”

Trump readying ‘all of the above’ fusillade on Chinese tech

The new visa policy shifts come as Trump is knee-deep in a controversial new tariff plan targeting Chinese trade and is intended to protect against the theft of U.S. intellectual property, or so the reasoning goes.

The visa change was signaled in the National Security Strategy report that the Trump administration issued in December. That document explains the rationale clearly:

The United States will review visa procedures to reduce economic theft by non-traditional intelligence collectors. We will consider restrictions on foreign STEM students from designated countries to ensure that intellectual property is not transferred to our competitors, while acknowledging the importance of recruiting the most advanced technical workforce to the United States.

The State Department noted these changes will go into effect starting on June 11.

Posted on

HoloLens acts as eyes for blind users and guides them with audio prompts

Microsoft’s HoloLens has an impressive ability to quickly sense its surroundings, but limiting it to displaying emails or game characters on them would show a lack of creativity. New research shows that it works quite well as a visual prosthesis for the vision impaired, not relaying actual visual data but guiding them in real time with audio cues and instructions.

The researchers, from CalTech and University of Southern California, first argue that restoring vision is at present simply not a realistic goal, but that replacing the perception portion of vision isn’t necessary to replicate the practical portion. After all, if you can tell where a chair is, you don’t need to see it to avoid it, right?

Crunching visual data and producing a map of high-level features like walls, obstacles, and doors is one of the core capabilities of the HoloLens, so the team decided to to let it do its thing and recreate the environment for the user from these extracted features.

They designed the system around sound, naturally. Every major object and feature can tell the user where it is, either via voice or sound. Walls, for instance, hiss (presumably a white noise, not a snake hiss) as the user approaches them. And the user can scan the scene, with objects announcing themselves from left to right from the direction in which they are located. A single object can be selected and will repeat its callout to help the user find it.

That’s all well for stationary tasks like finding your cane or the couch in a friend’s house. But the system also works in motion.

The team recruited seven blind people to test it out. They were given a brief intro but no training, and then asked to accomplish a variety of tasks. The users could reliably locate and point to objects from audio cues, and were able to find a chair in a room in a fraction of the time they normally would, and avoid obstacles easily as well.

This render shows the actual paths taken by the users in the navigation tests.

Then they were tasked with navigating from the entrance of a building to a room on the second floor by following the headset’s instructions. A “virtual guide” repeatedly says “follow me” from an apparent distance of a few feet ahead, while also warning when stairs were coming, where handrails were, and when the user had gone off course.

All seven users got to their destinations on the first try, and much more quickly than if they had had to proceed normally with no navigation. One subject, the paper notes, said “That was fun! When can I get one?”

Microsoft actually looked into something like this years ago, but the hardware just wasn’t there — HoloLens changes that. Even though it is clearly intended for use by sighted people, its capabilities naturally fill the requirements for a visual prosthesis like the one described here.

Interestingly, the researchers point out that this type of system was also predicted more than 30 years ago, long before they were even close to possible:

“I strongly believe that we should take a more sophisticated approach, utilizing the power of artificial intelligence for processing large amounts of detailed visual information in order to substitute for the missing functions of the eye and much of the visual pre-processing performed by the brain,” wrote the clearly far-sighted C.C. Collins way back in 1985.

The potential for a system like this is huge, but this is just a prototype. As systems like HoloLens get lighter and more powerful, they’ll go from lab-bound oddities to everyday items — one can imagine the front desk at a hotel or mall stocking a few to give to vision-impaired folks who need to find their room or a certain store.

“By this point we expect that the reader already has proposals in mind for enhancing the cognitive prosthesis,” they write. “A hardware/software platform is now available to rapidly implement those ideas and test them with human subjects. We hope that this will inspire developments to enhance perception for both blind and sighted people, using augmented auditory reality to communicate things that we cannot see.”

Posted on

Navigating the risks of artificial intelligence and machine learning in low-income countries

Aubra Anthony Contributor
Aubra Anthony is the strategy and research lead for the Center for Digital Development within the US Agency for International Development.

On a recent work trip, I found myself in a swanky-but-still-hip office of a private tech firm. I was drinking a freshly frothed cappuccino, eyeing a mini-fridge stocked with local beer, and standing amidst a group of hoodie-clad software developers typing away diligently at their laptops against a backdrop of Star Wars and xkcd comic wallpaper.

I wasn’t in Silicon Valley: I was in Johannesburg, South Africa, meeting with a firm that is designing machine learning (ML) tools for a local project backed by the U.S. Agency for International Development.

Around the world, tech startups are partnering with NGOs to bring machine learning and artificial intelligence (AI) to bear on problems that the international aid sector has wrestled with for decades. ML is uncovering new ways to increase crop yields for rural farmers. Computer vision lets us leverage aerial imagery to improve crisis relief efforts. Natural language processing helps usgauge community sentiment in poorly connected areas. I’m excited about what might come from all of this. I’m also worried.

AI and ML have huge promise, but they also have limitations. By nature, they learn from and mimic the status quo–whether or not that status quo is fair or just. We’ve seen AI or ML’s potential to hard-wire or amplify discrimination, exclude minorities, or just be rolled out without appropriate safeguards–so we know we should approach these tools with caution. Otherwise, we risk these technologies harming local communities, instead of being engines of progress.

Seemingly benign technical design choices can have far-reaching consequences. In model development, tradeoffs are everywhere. Some are obvious and easily quantifiable — like choosing to optimize a model for speed vs. precision. Sometimes it’s less clear. How you segment data or choose an output variable, for example, may affect predictive fairness across different sub-populations. You could end up tuning a model to excel for the majority while failing for a minority group.

Image courtesy of Getty Images

These issues matter whether you’re working in Silicon Valley or South Africa, but they’re exacerbated in low-income countries. There is often limited local AI expertise to tap into, and the tools’ more troubling aspects can be compounded by histories of ethnic conflict or systemic exclusion. Based on ongoing research and interviews with aid workers and technology firms, we’ve learned five basic things to keep in mind when applying AI and ML in low-income countries:

  1. Ask who’s not at the table. Often, the people who build the technology are culturally or geographically removed from their customers. This can lead to user-experience failures like Alexa misunderstanding a person’s accent. Or worse. Distant designers may be ill-equipped to spot problems with fairness or representation. A good rule of thumb: if everyone involved in your project has a lot in common with you, then you should probably work hard to bring in new, local voices.
  2. Let other people check your work. Not everyone defines fairness the same way, and even really smart people have blind spots. If you share your training data, design to enable external auditing, or plan for online testing, you’ll help advance the field by providing an example of how to do things right. You’ll also share risk more broadly and better manage your own ignorance. In the end, you’ll probably end up building something that works better.
  3. Doubt your data. A lot of AI conversations assume that we’re swimming in data. In places like the U.S., this might be true. In other countries, it isn’t even close. As of 2017, less than a third of Africa’s 1.25 billion people were online. If you want to use online behavior to learn about Africans’ political views or tastes in cinema, your sample will be disproportionately urban, male, and wealthy. Generalize from there and you’re likely to run into trouble.
  4. Respect context. A model developed for a particular application may fail catastrophically when taken out of its original context. So pay attention to how things change in different use cases or regions. That may just mean retraining a classifier to recognize new types of buildings, or it could mean challenging ingrained assumptions about human behavior.
  5. Automate with care. Keeping humans ‘in the loop’ can slow things down, but their mental models are more nuanced and flexible than your algorithm. Especially when deploying in an unfamiliar environment, it’s safer to take baby steps and make sure things are working the way you thought they would. A poorly-vetted tool can do real harm to real people.

AI and ML are still finding their footing in emerging markets. We have the chance to thoughtfully construct how we build these tools into our work so that fairness, transparency, and a recognition of our own ignorance are part of our process from day one. Otherwise, we may ultimately alienate or harm people who are already at the margins.

The developers I met in South Africa have embraced these concepts. Their work with the non-profit Harambee Youth Employment Accelerator has been structured to balance the perspectives of both the coders and those with deep local expertise in youth unemployment; the software developers are even foregoing time at their hip offices to code alongside Harambee’s team. They’ve prioritized inclusivity and context, and they’re approaching the tools with healthy, methodical skepticism. Harambee clearly recognizes the potential of machine learning to help address youth unemployment in South Africa–and they also recognize how critical it is to ‘get it right’. Here’s hoping that trend catches on with other global startups too.