24 April 2019

Evaluating the Unsupervised Learning of Disentangled Representations




The ability to understand high-dimensional data, and to distill that knowledge into useful representations in an unsupervised manner, remains a key challenge in deep learning. One approach to solving these challenges is through disentangled representations, models that capture the independent features of a given scene in such a way that if one feature changes, the others remain unaffected. If done successfully, a machine learning system that is designed to navigate the real world, such as a self driving car or a robot, can disentangle the different factors and properties of objects and their surroundings, enabling the generalization of knowledge to previously unobserved situations. While, unsupervised disentanglement methods have already been used for curiosity driven exploration, abstract reasoning, visual concept learning and domain adaptation for reinforcement learning, recent progress in the field makes it difficult to know how well different approaches work and the extent of their limitations.

In "Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations" (to appear at ICML 2019), we perform a large-scale evaluation on recent unsupervised disentanglement methods, challenging some common assumptions in order to suggest several improvements to future work on disentanglement learning. This evaluation is the result of training more than 12,000 models covering most prominent methods and evaluation metrics in a reproducible large-scale experimental study on seven different data sets. Importantly, we have also released both the code used in this study as well as more than 10,000 pretrained disentanglement models. The resulting library, disentanglement_lib, allows researchers to bootstrap their own research in this field and to easily replicate and verify our empirical results.

Understanding Disentanglement
To better understand the ground-truth properties of an image that can be encoded in a disentangled representation, first consider the ground-truth factors of the data set Shapes3D. In this toy model, shown in the figure below, each panel represents one factor that could be encoded into a vector representation of the image. The model shown is defined by the shape of the object in the middle of the image, its size, the rotation of the camera and the color of the floor, the wall and the object.
Visualization of the ground-truth factors of the Shapes3D data set: Floor color (upper left), wall color (upper middle), object color (upper right), object size (bottom left), object shape (bottom middle), and camera angle (bottom right).
The goal of disentangled representations is to build models that can capture these explanatory factors in a vector. The figure below presents a model with a 10-dimensional representation vector. Each of the 10 panels visualizes what information is captured in one of the 10 different coordinates of the representation. From the top right and the top middle panel we see that the model has successfully disentangled floor color, while the two bottom left panels indicate that object color and size are still entangled.
Visualization of the latent dimensions learned by a FactorVAE model (see below). The ground-truth factors wall and floor color as well as rotation of the camera are disentangled (see top right, top center and bottom center panels), while the ground-truth factors object shape, size and color are entangled (see top left and the two bottom left images).
Key Results of this Reproducible Large-scale Study
While the research community has proposed a variety of unsupervised approaches to learn disentangled representations based on variational autoencoders and has devised different metrics to quantify their level of disentanglement, to our knowledge no large-scale empirical study has evaluated these approaches in a unified manner. We propose a fair, reproducible experimental protocol to benchmark the state of unsupervised disentanglement learning by implementing six different state-of-the-art models (BetaVAE, AnnealedVAE, FactorVAE, DIP-VAE I/II and Beta-TCVAE) and six disentanglement metrics (BetaVAE score, FactorVAE score, MIG, SAP, Modularity and DCI Disentanglement). In total, we train and evaluate 12,800 such models on seven data sets. Key findings of our study include:
  • We do not find any empirical evidence that the considered models can be used to reliably learn disentangled representations in an unsupervised way, since random seeds and hyperparameters seem to matter more than the model choice. In other words, even if one trains a large number of models and some of them are disentangled, these disentangled representations seemingly cannot be identified without access to ground-truth labels. Furthermore, good hyperparameter values do not appear to consistently transfer across the data sets in our study. These results are consistent with the theorem we present in the paper, which states that the unsupervised learning of disentangled representations is impossible without inductive biases on both the data set and the models (i.e., one has to make assumptions about the data set and incorporate those assumptions into the model).
  • For the considered models and data sets, we cannot validate the assumption that disentanglement is useful for downstream tasks, e.g., that with disentangled representations it is possible to learn with fewer labeled observations.
The figure below demonstrates some of these findings. The choice of random seed across different runs has a larger impact on disentanglement scores than the model choice and the strength of regularization (while naively one might expect that more regularization should always lead to more disentanglement). A good run with a bad hyperparameter can easily beat a bad run with a good hyperparameter.
The violin plots show the distribution of FactorVAE scores attained by different models on the Cars3D data set. The left plot shows how the distribution changes as different disentanglement models are considered while the right plot displays the different distributions as the regularization strength in a FactorVAE model is varied. The key observation is that the violin plots substantially overlap which indicates that all methods strongly depend on the random seed.
Based on these results, we make four observations relevant to future research:
  1. Given the theoretical result that the unsupervised learning of disentangled representations without inductive biases is impossible, future work should clearly describe the imposed inductive biases and the role of both implicit and explicit supervision.
  2. Finding good inductive biases for unsupervised model selection that work across multiple data sets persists as a key open problem.
  3. The concrete practical benefits of enforcing a specific notion of disentanglement of the learned representations should be demonstrated. Promising directions include robotics, abstract reasoning and fairness.
  4. Experiments should be conducted in a reproducible experimental setup on a diverse selection of data sets.
Open Sourcing disentanglement_lib
In order for others to verify our results, we have released disentanglement_lib, the library we used to create the experimental study. It contains open-source implementations of the considered disentanglement methods and metrics, a standardized training and evaluation protocol, as well as visualization tools to better understand trained models.

The advantages of this library are three-fold. First, with less than four shell commands disentanglement_lib can be used to reproduce any of the models in our study. Second, researchers may easily modify our study to test additional hypotheses. Third, disentanglement_lib is easily extendible and can be used to bootstrap research into the learning of disentangled representations—it is easy to implement new models and compare them to our reference implementation using a fair, reproducible experimental setup.

Reproducing all the models in our study requires a computational effort of approximately 2.5 GPU years, which can be prohibitive. So, we have also released >10,000 pretrained disentanglement_lib models from our study that can be used together with disentanglement_lib.

We hope that this will accelerate research in this field by allowing other researchers to benchmark their new models against our pretrained models and to test new disentanglement metrics and visualization approaches on a diverse set of models.

Acknowledgments
This research was done in collaboration with Francesco Locatello, Mario Lucic, Stefan Bauer, Gunnar Rätsch, Sylvain Gelly and Bernhard Schöpf at Google AI Zürich, ETH Zürich and the Max-Planck Institute for Intelligent Systems. We also wish to thank Josip Djolonga, Ilya Tolstikhin, Michael Tschannen, Sjoerd van Steenkiste, Joan Puigcerver, Marcin Michalski, Marvin Ritter, Irina Higgins and the rest of the Google Brain team for helpful discussions, comments, technical help and code contributions.

It’s now easier to find work-from-home jobs in Google Search


Over the course of the last year or so, Google started adding more job search features to its search engine. Today, it’s expanding this program by also making it easier to find work-from-home jobs.

Typically, when you searched for a job like that, chances are you’ll be bombarded with a bunch of low-quality or even potentially malicious sites. Now, when you search for a job in Google Search using a query like ‘customer support jobs,’ you’ll be able to set the location to ‘work from home’ to find remote and telecommute jobs. Google says that a number of job listing sites, including Working Nomads, We Work Remotely and ZipRecruiter already support this feature. To filter these jobs, Google looks at the standard Schema.org markup that most job sites use to push their listings to Google Jobs.

“Through a number of user studies, we learned that most remote jobs were not clearly labeled as such,” Google product manager Jennifer Su writes in today’s announcement. “And our customers have told us it is difficult to accurately detect and return remote opportunities when users enter ‘work from home’ or related queries into their search box. By offering this capability to our customers, we hope to make remote work opportunities more discoverable to job seekers who need them.”

Google already lets job searchers filter by commute times and type of transit, so it makes sense to add this work-from-home option, too, especially now that more employees are looking for this kind of flexibility.


Read Full Article

Huawei’s P30 Pro excels on the camera front


It’s been a month since Huawei unveiled its latest flagship device — the Huawei P30 Pro. I’ve played with the P30 and P30 Pro for a few weeks and I’ve been impressed with the camera system.

The P30 Pro is the successor to the P20 Pro and features improvements across the board. It could have been a truly remarkable phone, but some issues still hold it back compared to more traditional Android phones, such as the Google Pixel 3 or OnePlus 6T.

A flagship device

The P30 Pro is by far the most premium device in the P line. It features a gigantic 6.47-inch OLED display, a small teardrop notch near the top, an integrated fingerprint sensor in the display and a lot of cameras.

Before diving into the camera system, let’s talk about the overall feel of the device. Compared to last year’s P20 Pro, the company removed the fingerprint sensor at the bottom of the screen and made the notch smaller. The fingerprint sensor doesn’t perform as well as a dedicated fingerprint sensor, but it gets the job done.

It has become hard to differentiate smartphones based on design as it looks a lot like the OnePlus 6T or the Samsung Galaxy S10. The display features a 19.5:9 aspect ratio with a 2340×1080 resolution, and it is curved around the edges.

The result is a phone with gentle curves. The industrial design is less angular, even though the top and bottom edges of the device have been flattened. Huawei uses an aluminum frame and a glass with colorful gradients on the back of the device.

Unfortunately, the curved display doesn’t work so well in practice. If you open an app with a unified white background, such as Gmail, you can see some odd-looking shadows near the edges.

Below the surface, the P30 Pro uses a Kirin 980 system-on-a-chip. Huawei’s homemade chip performs well. To be honest, smartphones have been performing well for a few years now. It’s hard to complain about performance anymore.

The phone features a headphone jack, a 40W USB-C charging port and an impressive 4,200 mAh battery. For the first time, Huawei added wireless charging to the P series (up to 15W).

You can also charge another phone or an accessory with reverse wireless charging, just like on the Samsung Galaxy S10. Unfortunately, you have to manually activate the feature in the settings every time you want to use it.

Huawei has also removed the speaker grill at the top of the display. The company now vibrates the screen in order to turn the screen into a tiny speaker for your calls. In my experience, it works well.

While the phone ships with Android Pie, Huawei still puts a lot of software customization with its EMUI user interface. There are a dozen useless Huawei apps that probably make sense in China, but don’t necessarily need to be there if you use Google apps.

For instance, the HiCare app keeps sending me notifications. The onboarding process is also quite confusing as some screens refer to Huawei features while others refer to standard Android features. It definitely won’t be a good experience for non tech-savvy people.


(P30 Pro on the left, P30 on the right)

Four cameras to rule them all

The P20 Pro already had some great camera sensors and paved the way for night photos in recent Android devices. The P30 Pro camera system can be summed up in two words — more and better.

The P30 Pro now features not one, not two, not three but f-o-u-r sensors on the back of the device.

  • The main camera is a 40 MP 27mm sensor with an f/1.6 aperture and optical image stabilization.
  • There’s a 20 MP ultra-wide angle lens (16mm) with an f/2.2 aperture.
  • The 8 MP telephoto lens provides nearly 5x optical zoom compared to the main lens (125mm) with an f/3.4 aperture and optical image stabilization.
  • There’s a new time-of-flight sensor below the flash of the P30 Pro. The phone projects infrared light and captures the reflection with this new sensor.

It has become a sort of a meme already — yes, the zoom works incredibly well on the P30 Pro. In addition to packing a lot of megapixels in the main sensor, the company added a telephoto lens with a periscope design. The sensor features a mirror to beam the light at a right angle and put more layers of glass in the sensor without making the phone too thick.

The company also combines the main camera sensor with the telephoto sensor to let you capture photos with a 10x zoom with a hybrid digital-optical zoom.

Here’s a photo series with the wide angle lens, the normal lens, a 5x zoom and a 10x zoom:

And it works incredibly well in daylight. Unfortunately, you won’t be able to use the telephoto lens at night as it doesn’t perform as well as the main camera.

In addition to hardware improvements, Huawei has also worked on the algorithms that process your shots. Night mode performs incredibly well. You just have to hold your phone for 8 seconds so that it can capture as much light as possible. Here’s what it looks like in a completely dark room vs. an iPhone X:

Huawei has also improved HDR processing and portrait photos. That new time-of-flight sensor works well when it comes to distinguishing a face from the background for instance.

Once again, Huawei is a bit too heavy-handed with post-processing. If you use your camera with the Master AI setting, colors are too saturated. The grass appears much greener than it is in reality. Skin smoothing with the selfie camera still feels weird too. The phone also aggressively smoothes surfaces on dark shots.

When you pick a smartphone brand, you also pick a certain photography style. I’m not a fan of saturated photos, so Huawei’s bias toward unnatural colors doesn’t work in my favor.

But if you like extremely vivid shots with insanely good sensors the P30 Pro is for you. That array of lenses also opens up a lot of possibilities and gives you more flexibility.

Fine prints

The P30 Pro isn’t available in the U.S. But the company has already covered the streets of major European cities with P30 Pro ads. It costs P30 Pro for €999 ($1,130) for 128GB of storage — there are more expensive options with more storage.

Huawei also unveiled a smaller device — the P30. It’s always interesting to look at the compromises of the more affordable model.

On that front, there’s a lot to like about the P30. For €799 ($900) with 128GB, you get a solid phone. It has a 6.1-inch OLED display and shares a lot of specifications with its bigger version.

The P30 features the same system-on-a-chip, the same teardrop notch, the same fingerprint sensor in the display, the same screen resolution. Surprisingly, the P30 Pro doesn’t have a headphone jack while the P30 has one.

There are some things you won’t find on the P30, such as wireless charging or the curved display. While the edges of the device are slightly curved, the display itself is completely flat. And I think it looks better.

Cameras are slightly worse on the P30, and you won’t be able to zoom in as aggressively. Here’s the full rundown:

  • A 40 MP main sensor with an f/1.8 aperture and optical image stabilization.
  • A 16 MP ultra-wide angle lens with an f/2.2 aperture.
  • An 8 MP telephoto lens that should provide 3x optical zoom.
  • No time-of-flight sensor.

In the end, it really depends what you’re looking for. The P30 Pro definitely has the best cameras of the P series. But the P30 is also an attractive phone for those looking for a smaller device.

Huawei has once again pushed the limits of what you can pack in a smartphone when it comes to cameras. While iOS and Android are more mature than ever, it’s fascinating to see that hardware improvements are not slowing down.


Read Full Article

Indian court lifts ban on TikTok in India


An Indian state court has reversed its ban on TikTok, allowing the short video app to return to both Apple and Google’s app stores, according to a report this morning from Reuters. Earlier this month, India’s Ministry of Electronics and Information Technology had ordered TikTok be removed from app stores, after a High Court in Madras determined the app was encouraging pornography and other illicit content.

Though the removal only affected new users who were looking to download TikTok’s app to their devices for the first time — not those who already had it installed — the ban was a major blow to TikTok’s Chinese owner Bytedance. The company said in a court filing the ban was resulting in a $500,000 daily loss, and was putting more than 250 jobs at risk.

India had become a large and growing market for TikTok, with nearly 300 million users in the country out of over 1 billion total downloads, according to Sensor Tower. (TikTok notes it had over 120 million monthly actives in India.)

India had also accounted for 27 percent of TikTok’s total installs between December 2017 and December 2018, Sensor Tower found, which meant the app was a huge source of TikTok’s overall growth.

However, some Indian politicians and parents believe the app’s content is inappropriate, particularly with regard to its use by minors. And the Tamil Nadu court — which ruled against TikTok — said the app could expose children to sexual predators, as well.

TikTok, meanwhile, had argued that a “very miniscule” proportion of its videos were inappropriate, and that it had removed over 6 million videos that had violated its terms of use and community guidelines after reviewing content created by users in India.

The ban, had it been upheld, could have foretold increased legal action and regulation against other social media apps in India.

This wasn’t the first time TikTok has come under fire by government regulators.

In February, the FTC in the U.S. fined TikTok $5.7 million for violating children’s privacy law (COPPA) and required the app to implement an age gate.

Bytedance, in a statement, welcomed the court’s decision to reverse the ban, saying:

We are glad about this decision and we believe it is also greatly welcomed by our thriving community in India, who use TikTok as a platform to showcase their creativity. We are grateful for the opportunity to continue serving our users better. While we’re pleased that our efforts to fight against misuse of the platform has been recognised, the work is never “done” on our end. We are committed to continuously enhancing our safety features as a testament to our ongoing commitment to our users in India


Read Full Article

Huawei’s P30 Pro excels on the camera front


It’s been a month since Huawei unveiled its latest flagship device — the Huawei P30 Pro. I’ve played with the P30 and P30 Pro for a few weeks and I’ve been impressed with the camera system.

The P30 Pro is the successor to the P20 Pro and features improvements across the board. It could have been a truly remarkable phone, but some issues still hold it back compared to more traditional Android phones, such as the Google Pixel 3 or OnePlus 6T.

A flagship device

The P30 Pro is by far the most premium device in the P line. It features a gigantic 6.47-inch OLED display, a small teardrop notch near the top, an integrated fingerprint sensor in the display and a lot of cameras.

Before diving into the camera system, let’s talk about the overall feel of the device. Compared to last year’s P20 Pro, the company removed the fingerprint sensor at the bottom of the screen and made the notch smaller. The fingerprint sensor doesn’t perform as well as a dedicated fingerprint sensor, but it gets the job done.

It has become hard to differentiate smartphones based on design as it looks a lot like the OnePlus 6T or the Samsung Galaxy S10. The display features a 19.5:9 aspect ratio with a 2340×1080 resolution, and it is curved around the edges.

The result is a phone with gentle curves. The industrial design is less angular, even though the top and bottom edges of the device have been flattened. Huawei uses an aluminum frame and a glass with colorful gradients on the back of the device.

Unfortunately, the curved display doesn’t work so well in practice. If you open an app with a unified white background, such as Gmail, you can see some odd-looking shadows near the edges.

Below the surface, the P30 Pro uses a Kirin 980 system-on-a-chip. Huawei’s homemade chip performs well. To be honest, smartphones have been performing well for a few years now. It’s hard to complain about performance anymore.

The phone features a headphone jack, a 40W USB-C charging port, an impressive 4,200 mAh battery. For the first time, Huawei added wireless charging to the P series (up to 15W).

You can also charge another phone or an accessory with reverse wireless charging, just like on the Samsung Galaxy S10. Unfortunately, you have to manually activate the feature in the settings every time you want to use it.

Huawei has also removed the speaker grill at the top of the display. The company now vibrates the screen in order to turn the screen into a tiny speaker for your calls. In my experience, it works well.

While the phone ships with Android Pie 9.1, Huawei still puts a lot of software customization with its EMUI user interface. There are a dozen useless Huawei apps that probably make sense in China, but don’t necessarily need to be there if you use Google apps.

For instance, the HiCare app keeps sending me notifications. The onboarding process is still also quite confusing as some screens refer to Huawei features while others refer to standard Android features. It definitely won’t be a good experience for non tech-savvy people.

Four cameras to rule them all

The P20 Pro already had some great camera sensors and paved the way for night photos in recent Android devices. The P30 Pro camera system can be summed up in two words — more and better.

The P30 Pro now features not one, not two, not three but f-o-u-r sensors on the back of the device.

  • The main camera is a 40 MP 27mm sensor with an f/1.6 aperture and optical image stabilization.
  • There’s a 20 MP ultra-wide angle lens (16mm) with an f/2.2 aperture.
  • The 8 MP telephoto lens provides nearly 5x optical zoom compared to the main lens (125mm) with an f/3.4 aperture and optical image stabilization.
  • There’s a new time-of-flight sensor below the flash of the P30 Pro. The phone projects infrared light and captures the reflection with this new sensor.

It has become a sort of a meme already — yes, the zoom works incredibly well on the P30 Pro. In addition to packing a lot of megapixels in the main sensor, the company added a telephoto lens with a periscope design. The sensor features a mirror to beam the light at a right angle and put more layers of glass in the sensor without making the phone too thick.

And it works incredibly well in daylight. Unfortunately, you won’t be able to use the telephoto lens at night as it doesn’t performa as well as the main camera.

The company also combines the main camera sensor with the telephoto sensor to let you capture photos with a 10x zoom with a hybrid digital-optical zoom.

In addition to hardware improvements, Huawei has also worked on the algorithms that process your shots. Night mode performs incredibly well. You just have to hold your phone for 8 seconds so that it can capture as much light as possible. Here’s what it looks like in a completely dark room vs. an iPhone X:

Huawei has also improved HDR processing and portrait photos. That new time-of-flight sensor works well when it comes to distinguishing a face from the background for instance.

Once again, Huawei is a bit too heavy-handed with post-processing. If you use your camera with the Master AI setting, colors are too saturated. The grass appears much greener than it is in reality. Skin smoothing with the selfie camera still feels weird too. The phone also aggressively smoothes surfaces on dark shots.

When you pick a smartphone brand, you also pick a certain photography style. I’m not a fan of saturated photos, so Huawei’s bias toward unnatural colors doesn’t work in my favor.

But if you like extremely vivid shots with insanely good sensors the P30 Pro is for you. That array of lenses also opens up a lot of possibilities and gives you more flexibility.

Fine prints

The P30 Pro isn’t available in the U.S. But the company has already covered the streets of major European cities with P30 Pro ads. It costs P30 Pro for €999 ($1,130) for 128GB of storage — there are more expensive options with more storage.

Huawei also unveiled a smaller device — the P30. It’s always interesting to look at the compromises of the more affordable model.

On that front, there’s a lot to like about the P30. For €799 ($900) with 128GB, you get a solid phone. It has a 6.1-inch OLED display and shares a lot of specifications with its bigger version.

The P30 features the same system-on-a-chip, the same teardrop notch, the same fingerprint sensor in the display, the same screen resolution. Surprisingly, the P30 Pro doesn’t have a headphone jack while the P30 has one.

There are some things you won’t find on the P30, such as wireless charging or the curved display. While the edges of the device are slightly curved, the display itself is completely flat. And I think it looks better.

Cameras are slightly worse on the P30, and you won’t be able to zoom in as aggressively. Here’s the full rundown:

  • A 40 MP main sensor with an f/1.8 aperture and optical image stabilization.
  • A 16 MP ultra-wide angle lens with an f/2.2 aperture.
  • An 8 MP telephoto lens that should provide 3x optical zoom.
  • No time-of-flight sensor.

In the end, it really depends what you’re looking for. The P30 Pro definitely has the best cameras of the P series. But the P30 is also an attractive phone for those looking for a smaller device.

Huawei has once again pushed the limits of what you can pack in a smartphone when it comes to cameras. While iOS and Android are more mature than ever, it’s fascinating to see that hardware improvements are not slowing down.


Read Full Article

Samsung Galaxy Fold teardown reveals ‘alarmingly fragile’ display


While Samsung performs its own internal investigations, the folks at iFixit had plenty to say about the Galaxy Fold in their own teardown. The writeup is a bit of a rollercoaster ride, admiring the “ambitious first-generation device,” while noting some clear flaws in the way the foldable was put together.

There’s a full 19 step breakdown of the product that adheres to the site’s customarily thorough breakdown, but the key takeaway here is the “alarmingly fragile” display mechanism. “Alarmingly” isn’t the kind of word you see tossed around lightly on a site like iFixit.

“Initial findings from the inspection of reported issues on the display showed that they could be associated with impact on the top and bottom exposed areas of the hinge,” Samsung wrote on Monday when it official pushed the Fold’s release back to an undisclosed date. That certainly appears to confirm the relatively fragile nature.

iFixit also did the unthinkable, pealing back the protective display, which has been mistaken for the user replaceable adhesive plastic Samsung’s other devices ship with. As expected, it didn’t go great. “In all known cases (including ours!), removing this layer kills the display,” the site writes. “The display could technically function without the layer, but it is so tightly adhered and the display is so fragile that it’s difficult to remove without applying display-breaking pressure.”

The site was suitably impressed with the hinge mechanism, but notes that “large gaps around the spine let dirt right in, possibly getting trapped between hinge and display.” That’s in line with one of the two points in Samsung’s own reporting of “an instance where substances found inside the device affected the display performance.”

The teardown appears to confirm fears that the issues could ultimately be broader than a few defective review units. Hopefully Samsung will push things back enough to properly address the above.


Read Full Article

Google Fit comes to iOS


Google today announced that Google Fit, the company’s fitness tracking app that launched on Android back in 2014, is now available on iOS.

It definitely took Google a while to bring the app to iOS. Until today, the only way to get your Fit data on your iPhone was in a special section of the Wear OS app on the iPhone. Without a Wear OS device, though, that section would’ve been empty.

If you’ve seen the Fit app on Android, then the iOS version will look very familiar. It features the same focus on Move Minutes and Heart Point, as well as the ability to pick up different activities based on your movement. You can also connect the app with apps connected to Apple Health like Sleep Cycle, Nike Run Club or Headspace can also sync with Google Fit.

Indeed, as a Google spokesperson told me, all of the movement data in the app also comes from Apple’s Health app — or from a Wear OS smartwatch, though few iOS users have opted to cross streams and use a Wear OS watch with their iPhones.

Since Apple Health already tracks your movement data, I’m not sure all that many iOS users will make the switch to Google Fit. It’s still good to see Google bring its service to this competing platform for those who maybe use multiple devices


Read Full Article

Google Fit comes to iOS


Google today announced that Google Fit, the company’s fitness tracking app that launched on Android back in 2014, is now available on iOS.

It definitely took Google a while to bring the app to iOS. Until today, the only way to get your Fit data on your iPhone was in a special section of the Wear OS app on the iPhone. Without a Wear OS device, though, that section would’ve been empty.

If you’ve seen the Fit app on Android, then the iOS version will look very familiar. It features the same focus on Move Minutes and Heart Point, as well as the ability to pick up different activities based on your movement. You can also connect the app with apps connected to Apple Health like Sleep Cycle, Nike Run Club or Headspace can also sync with Google Fit.

Indeed, as a Google spokesperson told me, all of the movement data in the app also comes from Apple’s Health app — or from a Wear OS smartwatch, though few iOS users have opted to cross streams and use a Wear OS watch with their iPhones.

Since Apple Health already tracks your movement data, I’m not sure all that many iOS users will make the switch to Google Fit. It’s still good to see Google bring its service to this competing platform for those who maybe use multiple devices


Read Full Article

Pew: U.S. adult Twitter users tend to be younger, more demographic; 10% create 80% of tweets


A new report out this morning from Pew Research Center offers insight into the U.S. adult Twitter population. The firm’s research indicates the Twitterverse tends to skew younger and more Democratic than the general public. It also notes that the activity on Twitter is dominated by a small percentage — most users rarely tweet, while the most prolific 10 percent are responsible for 80 percent of tweets from U.S. adults.

Pew says only around 22 percent of American adults today use Twitter, and they are representative of the broader population in some ways, but not in others.

For starters, Twitter’s U.S. adult users tend to be younger.

The study found the median age of Twitter users is 40, compared with the median age of U.S. adults, which is 47. Though less pronounced than the age differences, Twitter users also tend to have higher levels of household income and educational attainment, compared with the general population.

42 percent of adult Twitter users in the U.S. have at least a bachelor’s degree, which is 11 percentage points higher than the share of the public with this level of education (31%). Likely related to this is a higher income level. 41 percent of Twitter users have a household income above $75,000, which is 9 points higher than the same figure in the general population (32%).

A major difference — and a notable one, given yesterday’s sit-down between Twitter CEO Jack Dorsey and President Trump — is Pew’s discovery that 36 percent of Twitter U.S. adult users identify with the Democratic Party, versus 30 percent of U.S. adults (the latter, as per a November 2018 survey). Meanwhile, 21 percent of Twitter users identify as Republicans, versus 26 percent of U.S. adults. Political independents make up 29 percent of Twitter users, and a similar 27 percent of the general population.

Despite these differences, there are areas where Twitter users are more like the general U.S. adult population — specifically, in terms of the gender and racial makeup, Pew says.

In addition to the makeup of the adult population on Twitter, Pew also researched the activity on the platform, and found that the median user only tweets twice per month.

That means the conversation on Twitter is dominated by extremely active (or, in their parlance, “extremely online“) users. That means a large majority of Twitter’s content is created by a small number — 10 percent of users are responsible for 80 percent of all tweets from U.S. adults on Twitter.

The median user in this top 10 percent creates 138 tweets per month, favorites 70 posts per month, follows 456 accounts and has 387 followers. They tend to be women (65% are), and tend to tweet about politics (69% say they do.) They also more often use automated methods to tweet (25% do).

Meanwhile, the median users in the bottom 90 percent creates 2 tweets per month, favorites 1 post per month, follows 74 accounts, and has 19 followers. 48 percent are women, and 39 percent tweet about politics. Only 13 percent say they tweeted about politics in the last 30 days, compared with 42 percent of the top 10 percent of users. They are also less likely to use automated methods of tweeting, as only 15 percent do.

These differences lead to other ways where how the Twitterverse feels about key issues — like equality or immigration — differs from the general public, with viewpoints that lean more Democratic.

It’s worth noting, too, how the small amount of activity from a large group of Twitter users also speaks to Twitter’s inability to grow its monthly active user base (MAUs).

This week, Twitter reported its first quarter earnings and noted that its MAUs were 330 million in Q1, down by 6 million users from a year ago. Twitter now prefers to report on its monetizable daily active users — a metric that favors the app’s heavier users.

Pew’s research was conducted Nov. 21, 2018 through Dec. 17, 2018, among 2,791 U.S. adult Twitter users. The full report is available from Pew’s website.

 

 

 


Read Full Article

Twitter to offer report option for misleading election tweets


Twitter is adding a dedicated report option that enables users to tell it about misleading tweets related to voting — starting with elections taking place in India and the European Union.

From tomorrow users in India can report tweets they believe are trying to mislead voters — such as disinformation related to the date or location of polling stations; or fake claims about identity requirements for being able to vote — by tapping on the arrow menu of the suspicious tweet and selecting the ‘report tweet’ option then choosing: ‘It’s misleading about voting’.

Twitter says the tool will go live for the Indian Lok Sabha elections from tomorrow, and will launch in all European Union member states on April 29 — ahead of elections for the EU parliament next month.

The ‘misleading to voters’ option will persist in the list of available choices for reporting tweets for seven days after each election ends, Twitter said in a blog post announcing the feature.

In recent months the European Commission has been ramping up pressure on tech platforms to scrub disinformation ahead of elections to the EU parliament — issuing monthly reports on progress, or, well, the lack of it.

This follows a Commission initiative last year which saw major tech and ad platforms sign up to a voluntary Code of Practice on disinformation — including Facebook, Google and Twitter — committing themselves to take some non-prescribed actions to disrupt the ad revenues of disinformation agents and make political ads more transparent on their platform.

Another strand of the Code looks to have contributed to the development of Twitter’s new ‘misleading to voters’ report option — with signatories committing to:

  • Empower consumers to report disinformation and access different news sources, while improving the visibility and findability of authoritative content;

The EC says it will assess the Code’s initial 12-month period by the end of 2019 — and take a view on whether it needs to step in and propose regulation to control online disinformation. (Something which some individual EU Member States are already doing.)

In the latest progress report on the Code, which was published by the Commission yesterday but covers steps taken by the platforms in March 2019, it noted some progress made — but said it’s still not enough.

“Further technical improvements as well as sharing of methodology and data sets for fake accounts are necessary to allow third-party experts, fact-checkers and researchers to carry out independent evaluation,” EC commissioners warned in a joint statement.

In the case of Twitter the company was commended for having made political ad libraries publicly accessible but criticized (along with Google) for not doing more to improve transparency around issue-based advertising.

“It is regrettable that Google and Twitter have not yet reported further progress regarding transparency of issue-based advertising, meaning issues that are sources of important debate during elections,” the Commission said. 

It also reported that Twitter had provided figures on actions undertaken against spam and fake accounts but had failed to explain how these actions relate to activity in the EU.

“Twitter did not report on any actions to improve the scrutiny of ad placements or provide any metrics with respect to its commitments in this area,” it also noted.

In its blog post announcing the new tweet report option Twitter said it intends to roll the vote-focused feature out to “other elections globally throughout the rest of the year”, without providing further detail on which elections and markets it will prioritize for getting the tool.

“Our teams have been trained and we recently enhanced our appeals process in the event that we make the wrong call,” it added.


Read Full Article

Twitter to offer report option for misleading election tweets


Twitter is adding a dedicated report option that enables users to tell it about misleading tweets related to voting — starting with elections taking place in India and the European Union.

From tomorrow users in India can report tweets they believe are trying to mislead voters — such as disinformation related to the date or location of polling stations; or fake claims about identity requirements for being able to vote — by tapping on the arrow menu of the suspicious tweet and selecting the ‘report tweet’ option then choosing: ‘It’s misleading about voting’.

Twitter says the tool will go live for the Indian Lok Sabha elections from tomorrow, and will launch in all European Union member states on April 29 — ahead of elections for the EU parliament next month.

The ‘misleading to voters’ option will persist in the list of available choices for reporting tweets for seven days after each election ends, Twitter said in a blog post announcing the feature.

In recent months the European Commission has been ramping up pressure on tech platforms to scrub disinformation ahead of elections to the EU parliament — issuing monthly reports on progress, or, well, the lack of it.

This follows a Commission initiative last year which saw major tech and ad platforms — including Facebook, Google and Twitter — sign up to a voluntary Code of Practice on disinformation, committing themselves to take some non-prescribed actions to disrupt the ad revenues of disinformation agents and make political ads more transparent on their platform.

Another strand of the Code looks to have contributed to the development of Twitter’s new ‘misleading to voters’ report option — with signatories committing to:

  • Empower consumers to report disinformation and access different news sources, while improving the visibility and findability of authoritative content;

The EC says it will assess the Code’s initial 12-month period by the end of 2019 — and take a view on whether it needs to step in and propose regulation to control online disinformation. (Something which some individual EU Member States are already doing.)

In the latest progress report on the Code, which was published by the Commission yesterday but covers steps taken by the platforms in March 2019, it noted some progress made — but said it’s still not enough.

“Further technical improvements as well as sharing of methodology and data sets for fake accounts are necessary to allow third-party experts, fact-checkers and researchers to carry out independent evaluation,” EC commissioners warned in a joint statement.

In the case of Twitter the company was commended for having made political ad libraries publicly accessible but criticized (along with Google) for not doing more to improve transparency around issue-based advertising.

“It is regrettable that Google and Twitter have not yet reported further progress regarding transparency of issue-based advertising, meaning issues that are sources of important debate during elections,” the Commission said. 

It also reported that Twitter had provided figures on actions undertaken against spam and fake accounts but had failed to explain how these actions relate to activity in the EU.

“Twitter did not report on any actions to improve the scrutiny of ad placements or provide any metrics with respect to its commitments in this area,” it also noted.

In its blog post announcing the new tweet report option Twitter said it intends to roll the vote-focused feature out to “other elections globally throughout the rest of the year”, without providing further detail on which elections and markets it will prioritize for getting the tool.

“Our teams have been trained and we recently enhanced our appeals process in the event that we make the wrong call,” it added.


Read Full Article

Ride-hailing firm Grab is losing its CTO


Grab is once again on the hunt for a CTO after Theo Vassilakis, the former Microsoft and Google executive who currently occupies the role, announced that he will leave the ride-hailing company this summer.

Vassilakis became Grab CTO in October 2017, ending a very long search to fill the job, but he explained in a LinkedIn post that he is leaving the Singapore-based firm, and Southeast Asia, for family reasons.

My family and I moved to Singapore in late 2017 when I joined Grab. Living and working in Southeast Asia has been an adventure that broadened our horizons and will always be in our hearts. Unfortunately, our personal circumstances have changed unexpectedly and we’ll need to spend most of our time outside the region — mostly in the south of China for the foreseeable future.

Following his exit on June 30, Vassilakis will remain an advisor to Grab, with a specific focus on “coaching our senior tech leaders and shepherding our ongoing AI and marketplace optimization efforts.” He said that he will be involved in finding and hiring his replacement.

While it will lack a ‘group CTO,’ Grab does have CTOs for its transport and financial business units — Mark Porter and Vikas Agrawal, respectively — while head of product and design Jerald Singh will be involved in filling the void. Grab’s first CTO was Wei Zhu, who is credited with creating Connect with Facebook, but he left in 2015 after just a year and later sued over alleged unpaid earnings.

Under Vassilakis’ leadership, Grab massively increased its tech presence. The company now has seven R&D offices — Bangalore, Beijing, Ho Chi Minh City, Jakarta, Kuala Lumpur, Seattle and Singapore — and it claims to have doubled its headcount in 2018. Grab said in December that it is projected to add a further 1,000 “tech roles” this year.

The company has also expanded from merely transportation services to on-demand services, apps from third-parties via its ‘platform’ strategy, and payments and financial services.

Grab has also become the largest tech company in Southeast Asia by some margin in the eyes of investors. The company was most recently valued at $14 billion when it raised nearly $1.5 billion from SoftBank’s Vision Fund in March. To date, Grab has raised in excess of $7.5 billion from investors, and there’s more to come. The company said earlier this month that it plans to pull in $2 billion more from investors this year to battle rival Go-Jek, make acquisitions and develop its ‘super app’ strategy.


Read Full Article

The Mac Terminal Commands Cheat Sheet


mac-terminal-cheatsheet

macOS is an intuitive operating system, so you don’t have to spend lot of time learning the basics; Knowing this, why should you learn and take advantage of the Unix command line available on your Mac? We have four good reasons:

  1. There are dozens of open source and freely available Unix-based apps. You don’t have to spend money on these.
  2. When you’re having difficulty searching for files in Spotlight, you can turn to Unix search tools. They’re way more powerful than Spotlight.
  3. You can manage files, folders, and file archives in an automated manner. Setting up a cron job will handle this automatically.
  4. It gives you more power and control over your system.

With so many Mac commands, it’s often difficult to remember and use them all. We’re here to help with a detailed cheat sheet of Mac Terminal commands you can use to unlock enhanced productivity on your system.

Launch the Terminal app from Applications > Utilities or search for it via Spotlight. Then you can get started with some of the powerful commands below.

The Mac Terminal Commands Cheat Sheet

Command Action
Shortcuts
Tab Auto-complete file and folder names
Ctrl + A Go to the beginning of the line you're currently typing on
Ctrl + E Go to the end of the line you're currently typing on
Ctrl + U Clear the line before the cursor
Ctrl + K Clear the line after the cursor
Ctrl + W Delete the word before the cursor
Ctrl + T Swap the last two characters before the cursor
Esc + T Swap the last two words before the cursor
Ctrl + L Clear the screen
Ctrl + C Kill whatever you're running
Ctrl + D Exit the current shell
Option + → Move cursor one word forward
Option + ← Move cursor one word backward
Ctrl + F Move cursor one character forward
Ctrl + B Move cursor one character backward
Ctrl + Y Paste whatever was cut by the last command
Ctrl + Z Puts whatever you're running into a suspended background process
Ctrl + _ Undo the last command
Basics
/ (Forward Slash) Top level directory
. (Single Period) Current directory
.. (Double Period) Parent directory
~ (Tilde) Home directory
sudo [command] Run command with the security privileges of the super user
nano [file] Opens the Terminal editor
open [file] Opens a file
[command] -h Get help about a command
man [command] Show the help manual of the command
Change Directory
cd Home directory
cd [folder] Change directory, e.g. cd Documents
cd ~ Home directory
cd/ Root of the drive
cd - Previous directory or folder you last browsed
pwd Show your working directory
cd.. Move up to the parent directory
cd../.. Move up two levels
List Directory Contents
ls Display the name of files and subdirectories in the directory
ls -C Force multi-column output of the listing
ls -a List all entries including those with .(period) and ..(double period)
ls -1 Output the list of files in one entry per line format
ls -F Display a / (slash) immediately after each path that is a directory, * (asterisk) after executable programs or scripts, and @ after a symbolic link
ls -S Sort files or entries by size
ls -l List in a long format. Includes file mode, owner and group name, date and time file was modified, pathname, and more
ls -lt List the files sorted by time modified (most recent first)
ls -lh Long listing with human readable file sizes in KB, MB, or GB
ls -lo List the file names with size, owner, and flags
ls -la List detailed directory contents, including hidden files
File Size and Disk Space
du List usage for each subdirectory and its contents
du -sh [folder] Human readable output of all files in a directory
du -s Display an entry for each specified file
du -sk* | sort -nr List files and folders, totaling the size including the subfolders. Replace sk* with sm* to list directories in MB
df -h Calculate your system's free disk space
df -H Calculate free disk space in powers of 1,000 (as opposed to 1,024)
File and Directory Management
mkdir <dir> Create new folder named <dir>
mkdir -p <dir>/<dir> Create nested folders
mkdir <dir1> <dir2> <dir3> Create several folders at once
mkdir "<dir>" Create a folder with a space in the filename
rmdir <dir> Delete a folder (only works on empty folders)
rm -R <dir> Delete a folder and its contents
touch <file> Create a new file without any extension
cp <file> <dir> Copy a file to the folder
cp <file> <newfile> Copy a file to the current folder
cp <file>~/<dir>/<newfile> Copy a file to the folder and rename the copied file
cp -R <dir> <"new dir"> Copy a folder to a new folder with spaces in the filename
cp -i <file><dir> Prompts you before copying a file with a warning overwrite message
cp <file1> <file2> <file3>/Users/<dir> Copy multiple files to a folder
rm <file> Delete a file (This deletes the file permanently; use with caution.)
rm -i <file> Delete a file only when you give confirmation
rm -f <file> Force removal without confirmation
rm <file1> <file2> <file3> Delete multiple files without any confirmation
mv <file> <newfilename> Move/rename
mv <file> <dir> Move a file to the folder, possibly by overwriting an existing file
mv -i <file> <dir> Optional -i flag to warn you before overwriting the file
mv *.png ~/<dir> Move all PNG files from current folder to a different folder
Command History
Ctrl + R Search through previously used commands
history n Shows the previous commands you've typed. Add a number to limit to the last n items
![value] Execute the last command typed that starts with a value
!! Execute the last command typed
Permissions
ls -ld Display the default permission for a home directory
ls -ld/<dir> Display the read, write, and access permission of a particular folder
chmod 755 <file> Change the permission of a file to 755
chmod -R 600 <dir> Change the permission of a folder (and its contents) to 600
chown <user>:<group> <file> Change the ownership of a file to user and group. Add -R to include folder contents
Processes
ps -ax Output currently running processes. Here, a shows processes from all users and x shows processes that are not connected with the Terminal
ps -aux Shows all the processes with %cpu, %mem, page in, PID, and command
top Display live information about currently running processes
top -ocpu -s 5 Display processes sorted by CPU usage, updating every 5 seconds
top -o rsize Sort top by memory usage
kill PID Quit process with ID <PID>. You'll see PID as a column in the Activity Monitor
ps -ax | grep <appname> Find a process by name or PID
Network
ping <host> Ping host and display status
whois <domain> Output whois info for a domain
curl -O <url/to/file> Download file via HTTP, HTTPS, or FTP
ssh <username>@<host> Establish SSH connection to <host> with user <username>
scp <file><user>@<host>:/remote/path Copy <file> to a remote <host>
Homebrew
brew doctor Check brew for potential problems
brew install <formula> Install a formula
brew uninstall <formula> Uninstall a formula
brew list List all the installed formulas
brew search Display available formulas for brewing
brew upgrade Upgrade all outdated and unpinned brews
brew update Fetch latest version of homebrew and formula
brew cleanup Remove older version of installed formula
brew tap homebrew/cask Tap the cask repository from GitHub
brew cask list List all installed casks
brew cask install <cask> Install the given cask
brew cask uninstall <cask> Uninstall the given cask
Search
find <dir> -name <"file"> Find all files named <file> inside <dir>. Use wildcards (*) to search for parts of filenames
grep "<text>" <file> Output all occurrences of <text> inside <file> (add -i for case insensitivity)
grep -rl "<text>" <dir> Search for all files containing <text> inside <dir>
Output
cat <file> Output the content of <file>
less <file> Output the contents of <file> using the less command that supports pagination and more
head <file> Output the first 10 lines of <file>
<cmd> > > <file> Appends the output of <cmd> to <file>
<cmd> > <file> Direct the output of <cmd> into <file>
<cmd1> | <cmd2> Direct the output of <cmd1> to <cmd2>

Next, Customize the Terminal

There are lot of commands in this cheat sheet. But you don’t have to learn all of them at once! Pick a few that integrate well with your workflow and save you the most time. Once you’ve mastered these commands, there’s still more to learn about the Terminal to enhance your experience with it.

For further reading, we’ve looked at how to customize the Mac Terminal and make it more useful.

Read the full article: The Mac Terminal Commands Cheat Sheet


Read Full Article