Technology

The U.S.-China trade battle enters an important new phase. The South China Morning Post is reporting that ChinaMinistry of Commerce will clear Qualcomm pending $44 billion acquisition of NXP Semiconductors. One independent source also conveyed the same news to TechCrunch, although there has been no official word from Qualcomm, NXP or China at time of publication.

That acquisition was expected to close months ago, but the Chinese government repeatedly delayed its assent to the deal as part of its ongoing fight with the Trump administration over the future of bilateral trade. Chinaministry remained the last competition authority worldwide pending to approve the deal, and presumably it will close rapidly now that antitrust review has been completed.

The news of the approval broke just as The Wall Street Journal reported that the White House has authorized $50 billion in tariffs on Chinese goods. The final list of goods that will be subject to the tariffs has not been released, although TechCrunch has done a data analysis on the last set of tariffs, which focused on aluminum and steel imports. Direct news from the White House is expected Friday.

There has been a studied response and counter-response between the two countries over trade the past year, as both Presidents Trump and Xi Jinping sought high ground over the spat. The most recent set of issues has concerned ZTE, which was offered a reprieve by President Trump only to have its fate brought to Congress for a decision this week.

In my analysis on ZTEpotential death sentence, I wrote this afternoon that:

Ironically — and to be clear on this view, I am not getting this from sources, but rather pointing out a unique strategy vector here — it might well be Qualcomm that uses its DC policy shop to try to save ZTE. Those lobbyists protected Qualcomm from a takeover by Broadcom earlier this year, and it could try to make the case to Congress that it will be irreparably damaged if legislators don&t back off their threats.

The timing of the approval for Qualcomm could come with an understanding that it help ZTE with its congressional woes. Qualcomm has already agreed to form a strategic partnership with Baidu in the interim around AI and deep learning, which one source said to me was part of a package of concessions offered to placate Beijing.

Without a doubt, the news will prove a rare bit of relief for Qualcomm, which has been buffeted by challenges over the past year, including its hostile takeover battle with Broadcom and ongoing patent lawsuits with some of its biggest customers like Apple. Shareholders are likely to be enthusiastic with the outcome, and the stock was up 3 percent in after-hours trading following the news.

The acquisition of NXP is expected to provide a new set of technologies and patents for Qualcomm, particularly in strategic growth spaces like automotive, where Qualcomm has been weak on its product side.

Itbecome almost second nature to accept that cameras everywhere — from streets, to museums and shops — are watching you, but now they may be able to communicate with you, as well. New technology from Purdue University computer science researchers has made this dystopian prospect a reality in a new paper published today. But, they argue, itsafer than you might think.

The system is called PHADE, which allows for something called &private human addressing,& where camera systems and individual cell phones can communicate without transmitting any personal data, like an IP or Mac address. Instead of using an IP or Mac address, the technology relies on motion patterns for the address code. That way, even if a hacker intercepts it, they won&t be able to access the personphysical location.

Imagine you&re strolling through a museum and an unfamiliar painting catches your eye. The docents are busy with a tour group far across the gallery and you didn&t pay extra for the clunky recorder and headphones for an audio tour. While pondering the brushwork you feel your phone buzz, and suddenly a detailed description of the artwork and its painter is in the palm of your hand.

To achieve this effect, researchers use an approach similar to the kind of directional audio experience you might find at theme parks. Through processing the live video data, the technology is able to identify the individual motion patterns of pedestrians and when they are within a pertinent range — say, in front of a painting. From there they can broadcast a packet of information linked to the motion address of the pedestrian. When the userphone identifies that the motion address matches their own, the message is received.

While this tech can be used to better inform the casual museum-goer, the researchers also believe it has a role in protecting pedestrians from crime in their area.

&Our system serves as a bridge to connect surveillance cameras and people,& He Wang, a co-creator of the technology and assistant professor of computer science, said in a statement. &[It can] be used by government agencies to enhance public safety [by deploying] cameras in high-crime or high-accident areas and warn[ing] specific users about potential threats, such as suspicious followers.&

While the benefits of an increasingly interconnected world are still being debated and critiqued daily, there might just be an upside to knowing a cameragot its eye on you.

If I show you a single picture of a room, you can tell me right away that therea table with a chair in front of it, they&re probably about the same size, about this far from each other, with the walls this far away — enough to draw a rough map of the room. Computer vision systems don&t have this intuitive understanding of space, but the latest research from DeepMind brings them closer than ever before.

The new paper from the Google -owned research outfit was published today in the journal Science (complete with news item). It details a system whereby a neural network, knowing practically nothing, can look at one or two static 2D images of a scene and reconstruct a reasonably accurate 3D representation of it. We&re not talking about going from snapshots to full 3D images (Facebookworking on that) but rather replicating the intuitive and space-conscious way that all humans view and analyze the world.

When I say it knows practically nothing, I don&t mean itjust some standard machine learning system. But most computer vision algorithms work via whatcalled supervised learning, in which they ingest a great deal of data thatbeen labeled by humans with the correct answers — for example, images with everything in them outlined and named.

This new system, on the other hand, has no such knowledge to draw on. It works entirely independently of any ideas of how to see the world as we do, like how objects& colors change toward their edges, how they get bigger and smaller as their distance changes and so on.

It works, roughly speaking, like this. One half of the system is its &representation& part, which can observe a given 3D scene from some angle, encoding it in a complex mathematical form called a vector. Then therethe &generative& part, which, based only on the vectors created earlier, predicts what a different part of the scene would look like.

(A video showing a bit more of how this works is available here.)

Think of it like someone handing you a couple of pictures of a room, then asking you to draw what you&d see if you were standing in a specific spot in it. Again, this is simple enough for us, but computers have no natural ability to do it; their sense of sight, if we can call it that, is extremely rudimentary and literal, and of course machines lack imagination.

Yet there are few better words that describe the ability to say whatbehind something when you can&t see it.

&It was not at all clear that a neural network could ever learn to create images in such a precise and controlled manner,& said lead author of the paper, Ali Eslami, in a release accompanying the paper. &However we found that sufficiently deep networks can learn about perspective, occlusion and lighting, without any human engineering. This was a super surprising finding.&

It also allows the system to accurately recreate a 3D object from a single viewpoint, such as the blocks shown here:

I&m not sure I could do that.

Obviously therenothing in any single observation to tell the system that some part of the blocks extends forever away from the camera. But it creates a plausible version of the block structure regardless that is accurate in every way. Adding one or two more observations requires the system to rectify multiple views, but results in an even better representation.

This kind of ability is critical for robots, especially because they have to navigate the real world by sensing it and reacting to what they see. With limited information, such as some important clue thattemporarily hidden from view, they can freeze up or make illogical choices. But with something like this in their robotic brains, they could make reasonable assumptions about, say, the layout of a room without having to ground-truth every inch.

&Although we need more data and faster hardware before we can deploy this new type of system in the real world,& Eslami said, &it takes us one step closer to understanding how we may build agents that learn by themselves.&