How humans and machines can work together to save the tuna
From sashimi to tartare, tuna is in demand year-round. More than half of the world’s high-grade fish supply comes from the central and western Pacific Ocean, a region that accounts for a $7 billion market. The popularity of tuna sustains small fishing communities in distant places like Palau. But it also threatens an aquatic population that has been dwindling at an alarming rate.
Over the past few years, unregulated fishing practices in tuna-rich Pacific regions have threatened to wipe out rich species like the bluefin and bigeye. The ecological disturbance has raised a red flag among conservationists who are now looking to artificial intelligence for a solution to keep Palauan fisheries in check.
A majority of the 20,000 people in the island country of Palau rely on marine life for their livelihoods. The economy of the nation, which is spread across 250 islands in the Western Pacific Ocean, depends on tuna fishing and deep-diving tourism. The latter, driven by the shark population, accounts for the country’s $85 million deep-sea tourism industry. That breaks down to $170,000 per year for every alive shark, which is why the country enforced a ban on shark fishing in 2009 to build a sanctuary for divers. Even so, sharks and turtles are often unintentionally, even illegally, hauled in as bycatch on longline fishing boats.
In 2014, the Palauan government took more aggressive measures to protect the rich but vulnerable marine life. It announced a revolutionary ban on export fishing across its territorial waters — 200 nautical miles of the ocean surrounding the cluster of islands. And while local fisheries were allowed to operate, it became mandatory to have observers on board every tuna longliner. The new regulation was intended to keep overfishing and poaching in check. It would make vessel operators accountable for their catch, but the high costs of dedicating a person to every ship has kept the law from being fully implemented. A staggering 98 percent of the longline vessels continue to operate without any oversight in the region.
Palau’s longline tuna fishery is small in comparison with others in the Pacific. But it boasts of some of the most coveted fish species in the world. As such it plays a significant role in the ecological system and economic stability of Palau. To maintain that delicate balance, The Nature Conservancy, a global nonprofit organization, funds scientific research and provides technological solutions that can help establish a sustainable fishing model for the country.
In February of this year, the environmental organization bought fishing rights for a year in the Pacific region to test practices that would reduce bycatch. The Conservancy also launched a pilot project that has now equipped 24 boats with cameras and GPS devices to electronically monitor the fishing activities. The data fills in crucial gaps in knowledge about the catch on the boats and it supports the work of the human observers who can review the footage from multiple trips without being out at sea for months on end.
“We didn’t know what was happening in our waters,” says Kalei Luii, a compliance officer for the Division of Oceanic Fisheries Management in Palau, who works with the observers. “Bigeye tuna is our most prized fish, and their numbers were going down. [This] technology is helping us efficiently and sustainably manage our tuna fishery.”
An observer on a longline fishing boat in Palau. Photo: Jonne Roriz, The Nature Conservancy.
While the data allows more coverage across Palau’s tuna fishery, the process is incredibly time-consuming and expensive. Usually, an observer painstakingly watches hours of raw footage, sometimes hundreds of hours, to find any discrepancies and bad players. To make this manual review a lot less tedious, the Conservancy is now looking to machine learning for a scalable solution.
The organization has launched a search for the best AI solution through a contest with Kaggle, a startup that has a wide network of data scientists who partake in machine learning competitions. Over the next four months, the participating teams will have access to hours of fishing footage, allowing them to develop algorithms for Palau’s tuna fishery. Similar to facial recognition techniques employed at companies like Facebook to tag friends in a picture, the winning algorithm will teach machines to tell the difference between a tuna and a shark.
“I’m not trying to invent the future,” Matt Merrifield, chief technology officer at the Nature Conservancy in California, said in a recent announcement. “This technology exists, but it isn’t typically applied to conservation problems. This will be the first application of machine learning and artificial intelligence on board a fishing boat.” At the end of the contest, next year, the winning algorithm will be applied to all the boats with electronic monitoring systems in the Pacific.
The idea behind monitoring Palau’s fishery, through a combination of raw footage and algorithms, is to hold the fishermen accountable. “We want to stop illegal, unreported and unregulated fishing that is causing our tuna stocks to crash,” Luii said via email. “[It is] killing our vulnerable species like sharks and turtles and robbing our economy and those of other Pacific island nations.”
Luii doesn’t go out to sea, but he relies on the technologies that are being brought into the sustainable model. While he has not worked with the algorithms just yet, he is well acquainted with the data that comes in. He collects the hard drives at the end of a trip and works with the observers to review the footage. He says he was first drawn to the project because of his interest in the technology, but he continues to work with it because it aids his work.
“It complements the existing but lacking longline observer program here in Palau,” he says. “I’ve gained more understanding about how these longline vessels operate, and now [we’re] able to verify whether or not they comply with our laws at sea.”
The Nature Conservancy claims that the system, even when it will be loaded with machine learning algorithms next year, is not designed to do away with human observers. Instead, it’s expected to take away some of the drudge from their jobs. While that tends to be a common pro-AI argument, the people on the ground seem to agree with the sentiment in this case.
“I’m proud to be contributing to the sustainable management of my people’s resources,” says Luii. “My motivation is to help the people of Palau and make sure this fishery is thriving for future generations.”
DraftKings and FanDuel make their merger official
It was speculated back in June and reports last month indicated that a deal was close. Today, daily fantasy sports sites DraftKings and FanDuel made their merger official. The two sides announced a “strategic merger” that’s expected to close in the second half of 2017, but financial details weren’t disclosed.
The two companies say that the merger will allow them to focus more on building new features and products thanks to “operational efficiencies and cost savings.” Those features include more variety in contests, loyalty programs, better social tools and more sports-related content. Of course, DraftKings and FanDuel also say that merging the two entities will “accelerate its path to profitability.”
Perhaps more importantly, with one company at the forefront of daily fantasy, the two sides say they’ll be able to work better with government officials on regulations. It’s no secret that both sides have had their issues with lawmakers in the states where they operate. The tipping point of the argument is that daily fantasy differs from gambling because it requires skill rather than being left to chance or good luck. Previous legal troubles include a $12 million settlement with New York for false advertising that involved both DraftKings and FanDuel.
In terms of logistics, the new company with have offices in both Boston and New York, the current homes of DraftKings and FanDuel, respectively. When the merger is completed, DraftKings CEO Jason Robins will serve in the same position for the new company while FanDuel CEO Nigel Eccles will become chairman of the board. That board will have three directors from each company and one independent director. Of course, the merger agreement is subject to regulatory approval.
Google’s DeepMind AI gets a few new tricks to learn faster
When it comes to machine learning, every performance gain is worth a bit of celebration. That’s particularly true for Google’s DeepMind division, which has already proven itself by beating a Go world champion, mimicking human speech and cutting down their server power bills. Now, the team has unveiled new “reinforcement learning” methods to speed up how the AI platform trains itself without being directly taught.
First off, DeepMind’s learning agent has a better grasp of controlling pixels on the screen. Google notes it’s “similar to how a baby might learn to control their hands by moving them and observing the movements.” By doing this, it can figure out the best way to get high scores and play games more efficiently. Additionally, the agent can now figure out rewards from a game based on past performance. “By learning on rewarding histories much more frequently, the agent can discover visual features predictive of reward much faster,” Google says. The company laid out the entire concept for the abilities in a paper, “Reinforcement Learning with Unsupervised Auxiliary Tasks.”
These skills, along with DeepMind’s previous Deep Reinforcement Learning methods, make up the group’s new UNREAL (UNsupervised REinforcement and Auxiliary Learning) agent. That’s a mouthful, but the big takeaway is that DeepMind is beginning to teach itself much like humans. The group describes the methods as being similar to the way animals dream about positive and negative events (though I wouldn’t really say DeepMind has learned how to “dream”).
In a 3D maze environment called Labryinth, Google says the UNREAL agent was able to learn stages around ten times faster. It has managed to achieve 87 percent of “expert human performance” in that game, and around nine times typical human performance in a bevy of Atari titles.
On the face of it, UNREAL should help DeepMind’s agents significantly. But we’ll have to wait and see if those performance gains can actually be used in scenarios beyond games.
Source: DeepMind
The best Wi-Fi Range extender
By David Murphy
This post was done in partnership with The Wirecutter, a buyer’s guide to the best technology. When readers choose to buy The Wirecutter’s independently chosen editorial picks, it may earn affiliate commissions that support its work. Read the full article here.
After spending 140 total hours researching Wi-Fi extenders and testing more than 20, we recommend the TP-Link AC1750 Wi-Fi Range Extender RE450 for most people who need to boost their Wi-Fi signal to part of a house or apartment. Though the RE450 costs around $100 at this writing, it’s worth that price, as it offers the best combination of range, speed, portability, and physical connections of any extender we tested.
Why you should (or shouldn’t) buy this
Before you invest in a Wi-Fi extender, first you should make sure your router is near the center of your living space and out in the open if possible, not hidden in a closet or behind a desk. If you’ve used the same wireless router for years, consider upgrading to a better one, such as our pick for the best Wi-Fi router. If you still can’t get a signal where you need it, running an Ethernet cable directly from your router to a Wi-Fi access point or a router set up as one will get you much better performance than a Wi-Fi extender. But if you can’t run Ethernet and parts of your home or apartment still don’t get a good Wi-Fi signal, a wireless extender can help.
How we picked and tested
We tested the final 13 contenders after looking at 39 extenders. Photo: David Murphy
A good extender, like a modern router, should support both the older, more crowded 2.4 GHz band and the newer, faster 5 GHz band. You also need a dual-band extender to avoid the performance hit of connecting to your devices on the same band as the one for connecting to your router. An extender should also support at least two spatial streams (also called data streams) on each band. We looked at any N600 or AC1200-plus extenders from major manufacturers that met these criteria, had good reviews (if any), and weren’t too expensive.
To test the 13 final contenders, we set up one of the fastest routers we’ve tested, Netgear’s R8500, in one corner of a 2,577-square-foot, single-story house, with the extender in the next room, to minimize the effects of the router’s performance on the extender’s benchmarks and to test the range of the extenders’ signals.
We connected each extender to the router’s 5 GHz Wi-Fi signal. For the tests, we connected an Asus ZenBook UX305LA (which uses two-stream 802.11ac Wi-Fi) to each extender from two locations—one obstructed and one unobstructed—both 43 feet away from the extender.
We tested the extenders using iPerf3, a network monitoring and measuring tool, to evaluate data transfers between a desktop PC (connected to the router via Gigabit Ethernet) and our test laptop. Each iPerf3 test attempted to transfer as much data as possible from the test laptop to the desktop PC (via a single TCP connection). We let 15 seconds elapse before recording the average transfer speed across 60 one-second intervals, and ran each test for each extender, on each band, at each test location. To learn more about our testing plan, read our full guide.
Our pick
The TP-Link RE450 will eat up most of your wall socket when you plug it in—the price you pay for great Wi-Fi range. Photo: David Murphy
The TP-Link AC1750 Wi-Fi Range Extender RE450 is the best wireless extender for most people because it offers incredible performance at long range and supports the fastest wireless speeds of any device you’re likely to own, even if you have a MacBook Pro. It was the only extender that hit triple-digit speeds on our easier long-range 5 GHz test, and its long-range 2.4 GHz performance was better than that of everything else we tested. This model is simple to set up, and it has a few useful features within its easy-to-navigate user interface.
The biggest issue with the RE450 is its size. It’s huge. If you plug it into a wall outlet, bigger devices (like a power strip) likely won’t have room.
Even so, in each of our tests, the RE450 had not just the best performance of any of the extenders in the group, but also the best price-to-performance ratio (aside from our slower, cheaper budget pick). You spend about as much for the RE450 as you would for our best Wi-Fi router pick, but you get proportionate speed and range—the most of any extender we’ve recently tested.
For home offices and entertainment centers
Linksys’s RE6500 isn’t as convenient as a wall-plug extender, but it has more Gigabit Ethernet ports and doesn’t take up much space on a coffee table or shelf. Photo: David Murphy
If our pick is sold out or unavailable, or if you have a lot of wired devices that need access to your home network, the Linksys AC1200 Max Wi-Fi Range Extender RE6500 is an excellent alternative. In our tests it gave us around three-fourths the speed of our primary pick; of all the wireless-ac extenders we recently tested, however, it had the second-fastest average speeds and the second-best price-to-performance ratio. In addition, it won’t block an extra wall outlet, it has three more Gigabit Ethernet ports than our pick, and its user interface is more helpful for people with less networking experience. Read more in our full guide.
The low-cost, last-gen alternative
The Asus RP-N53 is a tiny, wireless-n extender with a signal-strength LED indicator plus another LED on the back that works as a night-light. Photo: David Murphy
If you just need Wi-Fi in a far-flung area, don’t care about 802.11ac, and don’t want to spend a lot of money, get the Asus RP-N53 Dual-Band Wireless-N600 Range Extender. This wall-plug 802.11n extender is notably small, but it offers better performance and features than other extenders in its price range, including an easy-to-use Web configuration screen, music-streaming capabilities, and the ability to function as a night-light. To learn more, see our full guide.
This guide may have been updated by The Wirecutter. To see the current recommendation, please go here.
Note from The Wirecutter: When readers choose to buy our independently chosen editorial picks, we may earn affiliate commissions that support our work.
Lysol owner and Indiegogo team up to find the next… Lysol
Reckitt Benckiser’s board thinks that there might be something in this crowdfunding lark for sourcing new product ideas. That’s why the chemicals company behind Durex, Veet and Lysol is teaming up with Indiegogo for its Healthier Tomorrow Challenge. Money-hungry startups with ideas related to health products are offered the chance to take part in a Shark Tank-esque competition for RB’s love. If you have an idea that the conglomerate thinks is a winner, you’ll be offered support, mentoring and, potentially, access to RB’s manufacturing and distribution channels.
The competition is broken down into categories including alleviating flu symptoms, drug-free pain relief and sexual wellbeing tools. If you develop, for instance, a new long-distance sex toy, TENS machine or a wearable that reminds people to take their vitamins, then sign up. Although if your magical new product doesn’t fit into one of those categories, you’re still encouraged to sign up anyway. The partnership with Indiegogo is designed to encourage potential ideas to go to the market to see if people like them — a metric by which RB is likely to look favorably.
Source: Indiegogo
iHome Launching First Clock Radio That Can Charge Both an iPhone and Apple Watch
iHome has announced it will soon release the iPLWBT5, the first alarm clock radio featuring both Lightning and Apple Watch chargers.
Audio can be played via the docked Lightning device, Bluetooth, USB, or AUX. The dock features an alarm clock with dual alarms, FM radio with 6 preset stations, auto dimming display, built-in microphone, digital voice echo cancellation, answer/end controls for speakerphone calls, and a 1 amp USB port for charging devices.
iHome’s iPLWBT5 will be available in mid December for $129.99 online and at retailers such as Best Buy and Bed Bath & Beyond.
Related Roundups: Apple Watch Series 2, watchOS 3, iPhone 7
Tag: iHome
Buyer’s Guide: Apple Watch (Neutral)
Discuss this article in our forums
First HomeKit-Enabled ‘Smart Home Packages’ Open to Homebuyers in San Jose
Homebuilding company KB Home today announced “the nation’s first HomeKit-enabled community,” which comes with the opportunity for buyers to build homes with Apple’s HomeKit platform integrated from the get-go. A report last month brought news that Apple was working with homebuilders to install HomeKit and the Apple ecosystem of connected devices into properties at the beginning of construction.
Now the Promenade at Communications Hill in San Jose will be the first community of homes to include a HomeKit smart package that buyers can purchase. Following the debut in San Jose, the packages will also be available for curious home buyers at KB Home’s Magnolia at Patterson Ranch community.
A few exterior models of the homes in San Jose
As detailed in the original report, the smart homes utilize HomeKit at the beginning of a home’s construction to install products like lights, window blinds, garage doors, doorbells, and more at added cost, so the homeowners don’t have to add them gradually over time. In one of the test homes displayed in Alameda, California, a total of $30,000 worth of smart products fueled the smart package.
A model home showcasing how KB homeowners can simply and securely monitor and control Apple HomeKit-compatible accessories, such as lighting, door locks and other security features, and more, conveniently from their iPhone, iPad or Apple Watch, is now open at Promenade at Communications Hill.
Designed with today’s tech-savvy buyer in mind, Promenade at Communications Hill offers convenient city living close to San Jose’s shopping, dining, nightlife and cultural attractions, as well as to the area’s major Silicon Valley employers. The community also provides easy access to Interstates 880, 680 and U.S. Highway 101.
KB Home said that the plans available at the Promenade community range in size from 2,148 to 2,178 square feet, with the option to choose up to five bedrooms, four baths, and a two-car garage. Pricing on the properties begin in the low-$900,000s, which seemingly doesn’t include the smart package expenses.
Tag: HomeKit
Discuss this article in our forums
Apple May Be Throttling Verizon iPhone 7 LTE Performance
Apple may be throttling the LTE performance of the Verizon iPhone 7 to keep it on par with the AT&T iPhone 7, according to research conducted by Twin Prime and Cellular Insights and shared by Bloomberg. Based on testing, the Verizon iPhone 7 performs about as well as the AT&T iPhone 7, but it does not reach the data transfer speeds that it is capable of hitting.
The Verizon (and Sprint) iPhone 7 models use different LTE hardware than the AT&T (and T-Mobile) iPhone 7 models, adopting a Qualcomm LTE modem instead of an Intel LTE modem.
The hardware from Qualcomm is capable of maximum theoretical download speeds of 600Mb/s, while the Intel LTE modem tops out at 450Mb/s, but the Qualcomm-equipped Verizon iPhone 7 is only marginally outperforming the AT&T iPhone 7. Researchers believe Apple may be hobbling the Verizon iPhone 7 by not taking advantage of “a crucial component,” thereby ensuring all models of the iPhone 7 perform at a similar level.
“The data indicates that the iPhone 7 is not taking advantage of all of Verizon’s network capabilities,” said Gabriel Tavridis, head of product at Twin Prime. “I doubt that Apple is throttling each bit on the Verizon iPhone, but it could have chosen to not enable certain features of the network chip.” […]
Field tests suggest the Verizon iPhone 7 is just a “little faster” than the AT&T iPhone 7, but not as fast as it could be. Tests were conducted by comparing iPhone 7 performance on the Verizon network to a Samsung Galaxy S7, which also uses the Qualcomm X12. Data collected from more than 100,000 phones downloading the same image indicates the S7 was twice as fast as the iPhone 7.
In a statement, an Apple spokesperson said there is no discernible difference in the wireless performance of any of the iPhone 7 models.
“Every iPhone 7 and iPhone 7 Plus meets or exceeds all of Apple’s wireless performance standards, quality metrics, and reliability testing,” Apple spokeswoman Trudy Muller said. “In all of our rigorous lab tests based on wireless industry standards, in thousands of hours of real-world field testing, and in extensive carrier partner testing, the data shows there is no discernible difference in the wireless performance of any of the models.”
Though the two phones are performing on a similar level, past testing by Cellular Insights suggests that changes somewhat when signal strength is an issue. In areas of weak reception, the Verizon iPhone noticeably outperforms the AT&T iPhone by maintaining a stronger connection and enabling faster low-signal transfer speeds.
Other network testing companies that Bloomberg contacted said that reliably measuring data transfer speeds is difficult due to the many factors that can affect performance, but the information collected from Cellular Insights and Twin Prime was not disputed.
Related Roundup: iPhone 7
Discuss this article in our forums
Google Pixel vs. Galaxy S7 video stabilization test: Does OIS make a difference?
How does the Pixel’s crazy gyro-based stabilization compare to the Galaxy S7’s OIS?
The Samsung Galaxy S7 and Google Pixel take radically different approaches to stabilizing your videos. On the GS7, you’ve got traditional optical image stabilization — a hardware feature that cushions the camera module to make video more stable, while also helping out with still shots in darker conditions. In addition, there’s traditional software-based video stabilization available in the Settings menu.
On the Pixel, there’s no physical stabilization — instead, Google’s digital stabilization works with the phone’s gyroscope to stabilize footage, with surprisingly great results.
So how do these differing approaches compare? Let’s take a look.
Note: We tested both phones at Full HD (1080p) resolution at 30 frames per second, since that’s the default setting for both the GS7 and Pixel.
Walking
For our first test, we took the Pixel and Galaxy S7 out for a stroll. Both phones were held one-handed at a moderate walking pace at mid-day on an overcast day.
With stabilization turned off (the default for the GS7), Samsung’s phone shows the telltale “wobbling” of an OIS-equipped camera. It’s actually not too nauseating, but without any software stabilization going on more susceptible to sudden movements. That said, we’re getting a full 1080p image here with plenty of detail, unlike the two stabilized clips.
With stabilization enabled, the footage from the GS7 becomes smoother, but with a cropped-in frame and less fine detail. Sudden movements are still visible in places, but the footage is noticeably smoother.
The Pixel produces a less saturated video than the GS7, with dynamic range taking a small hit too (notice the washed-out sky), however footage is almost unnaturally smooth. Google’s phone even handles the motion of turning around halfway through the shot relatively well.
Panning
For our next test, we moved each phone through a figure-of-eight motion to see how well they handled smooth movement.
The Galaxy S7 without stabilization captured more fine detail than the other tests, but motion is noticeably jerky throughout. It’s pretty unpleasant to watch.
Bring software stabilization into the mix on the GS7 and there’s some noticeable ghosting and blurring, but otherwise the movement is smooth.
In this instance, the Pixel offers the best of both worlds — a wider viewport and smooth motion throughout, and nowhere near as much ghosting as the Galaxy S7.
Faster movement
For a more challenging test, we took the phones for a faster walk with dusk approaching. This time around, the Galaxy S7 (without stabilization) probably produced the best-looking footage, although with plenty of OIS-induced wobble.
With software stabilization enabled, the GS7 admirably evens out the jarring motion you see in the first clip, but there’s a ton of ghosting, which is probably even more distracting. On top of that, fine detail gets obliterated, leaving you with a very soft image.
The Pixel does slightly better, with more fine detail and less ghosting, but the overall image has a distracting shimmer to it.
Side-by-side comparison
Galaxy S7 and Pixel side-by-side, both with stabilization enabled. Check out the difference.
Low light
Here’s where things get interesting. As before, the GS7 without stabilization captures way more fine detail, and in this time around it’s easily the best-looking of the three videos, even with lots of motion.
Enabling stabilization on the GS7 in these conditions produces a soft, ghosty, unwatchable mess. (Notice the blurring around the streetlights and headlights.)
The video from the Pixel is still very soft, with not much fine detail at all, but it’s easier on the eyes than the stabilized GS7 footage. Colors are also more accurate, a hallmark of the Pixel when shooting in low light.
Wrapping things up
The Google Pixel manages to pull off some pretty impressive stabilization tricks, thanks to its use of data from the phone’s gyroscope. By contrast, the Galaxy S7’s stabilization mode (disabled by default) degrades image quality much more aggressively. That said, footage from the GS7 without stabilization enabled readily captures more fine detail than the Pixel, with slightly wider dynamic range.
A few conclusions based on our testing:
- Any kind of software stabilization, even as good as the Pixel’s, comes with some trade-off in terms of image quality.
- The Pixel’s digital stabilization is generally better than the Galaxy S7’s, thanks to its use of gyroscope data from the phone.
- Even with OIS, you’ll get smoother pans with some kind of software stabilization.
- The impact of digital stabilization becomes more noticeable in lower-light conditions. At night, you’ll lose a lot of fine detail.
- If you’re shooting on a Galaxy S7, you probably want to leave the “video stabilization” option turned off.
- On the Pixel, video stabilization is enabled by default, and you should probably leave it that way.
Pixel and GS7 owners, how have you found the phones’ video cameras? Let us know down in the comments!
Google Pixel + Pixel XL
- Google Pixel and Pixel XL review
- Google Pixel XL review: A U.S. perspective
- Google Pixel FAQ: Should you upgrade?
- Pixel + Pixel XL specs
- Understanding Android 7.1 Nougat
- Join the discussion in the forums!
Google Store
Verizon
Samsung Galaxy S7 and S7 edge
- Galaxy S7 review
- Galaxy S7 edge review
- U.S. unlocked Galaxy S7
- Should you upgrade to the Galaxy S7?
- Best SD cards for Galaxy S7
- Join our Galaxy S7 forums
Unlocked
AT&T
Sprint
T-Mobile
Verizon
AIVAnet 