Timesdelhi.com

December 12, 2018
Category archive

Artificial Intelligence

Krisp reduces noise on calls using machine learning, and it’s coming to Windows soon

in Artificial Intelligence/audio/Delhi/funding/Gadgets/India/noise reduction/Politics/TC by

If your luck is anything like mine, as soon as you jump on an important call, someone decides it’s a great time to blow some leaves off the sidewalk outside your window. 2Hz’s Krisp is a new desktop app that uses machine learning to subtract background noise like that, or crowds, or even crying kids — while keeping your voice intact. It’s already out for Macs and it’s coming to Windows soon.

I met the creators of Krisp, including 2Hz co-founder Davit Baghdasaryan, earlier this year at UC Berkeley’s Skydeck accelerator, where they demonstrated their then-prototype tech.

The tech involved is complex, but the idea is simple: If you create a machine learning system that understand what the human voice sounds like, on average, then it can listen to an audio signal and select only that part of it, cutting out a great deal of background noise.

Baghdasaryan, formerly of Twilio, originally wanted to create something that would run on mobile networks, so T-Mobile or whoever could tout built-in noise cancellation. This platform approach proved too slow, however, so they decided to go straight to consumers.

“Traction with customers was slow, and this was a problem for a young startup,” Baghdasaryan said in an email later. However, people were loving the idea of ‘muting noise,’ so we decided to switch all our focus and build a user-facing product.”

That was around the time I talked with them in person, incidentally, and just six months later they had released on Mac.

It’s simple: you run the app, and it modifies both the outgoing and incoming audio signals, with the normal noisy signal going in one end and a clean, voice-focused one coming out the other. Everything happens on-device and with very short latency (around 15 milliseconds), so there’s no cloud involved and nothing is ever sent to any server or even stored locally. The team is working on having the software adapt and learn on the fly, but it’s not implemented yet.

Another benefit of this approach is it doesn’t need any special tweaking to work with, say, Skype instead of Webex. Because it works at the level of the OS’s sound processing, whatever app you use just hears the Krisp-modified signal as if it were clean out of your mic.

They launched on Mac because they felt the early-adopter type was more likely to be on Apple’s platform, and the bet seems to have paid off. But a Windows version is coming soon — the exact date isn’t set, but expect it either late this month or early January. (We’ll let you know when it’s live.)

It should be more or less identical to the Mac version, but there will be a special gaming-focused one. Gamers, Baghdasaryan pointed out, are much more likely to have GPUs to run Krisp on, and also have a real need for clear communication (as a PUBG player I can speak to the annoyance of an open mic and clacky keys). So there will likely be a few power user features specific to gamers, but it’s not set in stone yet.

You may wonder, as I did, why they weren’t going after chip manufacturers, perhaps to include Krisp as a tech built into a phone or computer’s audio processor.

In person, they suggested that this ultimately was also too slow and restrictive. Meanwhile, they saw that there was no real competition in the software space, which is massively easier to enter.

“All current noise cancellation solutions require multiple microphones and a special form factor where the mouth must be close to one of the mics. We have no such requirement,” Baghdasaryan explained. “We can do it with single-mic or operate on an audio stream coming from the network. This makes it possible to run the software in any environment you want (edge or network) and any direction (inbound or outbound).”

If you’re curious about the technical side of things — how it was done with one mic, or at low latency, and so on — there’s a nice explanation Baghdasaryan wrote for the Nvidia blog a little while back.

Furthermore, a proliferation of AI-focused chips that Krisp can run on easily means easy entry to the mobile and embedded space. “We have already successfully ported our DNN to NVIDIA GPUs, Intel CPU/GNA, and ARM. Qualcomm is in the pipeline,” noted Baghdasaryan.

To pursue this work the company has raised a total of $2 million so far: $500K from Skydeck as well as friends and family for a pre-seed round, then a $1.5 M round led by Sierra Ventures and Shanda Group.

Expect the Windows release later this winter, and if you’re already a user, expect a few new features to come your way in the same time scale. You can download Krisp for free here.

News Source = techcrunch.com

Why you need a supercomputer to build a house

in affordable housing/Artificial Intelligence/building/building codes/buildings/camino/concur/concur labs/Cove.Tool/cover/Cover Technologies/Delhi/Developer/Enterprise/envelope/Government/GreenTech/housing/India/Logistics/machine learning/Policy/Politics/Real estate/regulation/SaaS/Startups/TC/zoning by

When the hell did building a house become so complicated?

Don’t let the folks on HGTV fool you. The process of building a home nowadays is incredibly painful. Just applying for the necessary permits can be a soul-crushing undertaking that’ll have you running around the city, filling out useless forms, and waiting in motionless lines under fluorescent lights at City Hall wondering whether you should have just moved back in with your parents.

Consider this an ongoing discussion about Urban Tech, its intersection with regulation, issues of public service, and other complexities that people have full PHDs on. I’m just a bitter, born-and-bred New Yorker trying to figure out why I’ve been stuck in between subway stops for the last 15 minutes, so please reach out with your take on any of these thoughts: @Arman.Tabatabai@techcrunch.com.

And to actually get approval for those permits, your future home will have to satisfy a set of conditions that is a factorial of complex and conflicting federal, state and city building codes, separate sets of fire and energy requirements, and quasi-legal construction standards set by various independent agencies.

It wasn’t always this hard – remember when you’d hear people say “my grandparents built this house with their bare hands?” These proliferating rules have been among the main causes of the rapidly rising cost of housing in America and other developed nations. The good news is that a new generation of startups is identifying and simplifying these thickets of rules, and the future of housing may be determined as much by machine learning as woodworking.

When directions become deterrents

Photo by Bill Oxford via Getty Images

Cities once solely created the building codes that dictate the requirements for almost every aspect of a building’s design, and they structured those guidelines based on local terrain, climates and risks. Over time, townships, states, federally-recognized organizations and independent groups that sprouted from the insurance industry further created their own “model” building codes.

The complexity starts here. The federal codes and independent agency standards are optional for states, who have their own codes which are optional for cities, who have their own codes that are often inconsistent with the state’s and are optional for individual townships. Thus, local building codes are these ever-changing and constantly-swelling mutant books made up of whichever aspects of these different codes local governments choose to mix together. For instance, New York City’s building code is made up of five sections, 76 chapters and 35 appendices, alongside a separate set of 67 updates (The 2014 edition is available as a book for $155, and it makes a great gift for someone you never want to talk to again).

In short: what a shit show.

Because of the hyper-localized and overlapping nature of building codes, a home in one location can be subject to a completely different set of requirements than one elsewhere. So it’s really freaking difficult to even understand what you’re allowed to build, the conditions you need to satisfy, and how to best meet those conditions.

There are certain levels of complexity in housing codes that are hard to avoid. The structural integrity of a home is dependent on everything from walls to erosion and wind-flow. There are countless types of material and technology used in buildings, all of which are constantly evolving.

Thus, each thousand-page codebook from the various federal, state, city, township and independent agencies – all dictating interconnecting, location and structure-dependent needs – lead to an incredibly expansive decision tree that requires an endless set of simulations to fully understand all the options you have to reach compliance, and their respective cost-effectiveness and efficiency.

So homebuilders are often forced to turn to costly consultants or settle on designs that satisfy code but aren’t cost-efficient. And if construction issues cause you to fall short of the outcomes you expected, you could face hefty fines, delays or gigantic cost overruns from redesigns and rebuilds. All these costs flow through the lifecycle of a building, ultimately impacting affordability and access for homeowners and renters.

Startups are helping people crack the code

Photo by Caiaimage/Rafal Rodzoch via Getty Images

Strap on your hard hat – there may be hope for your dream home after all.

The friction, inefficiencies, and pure agony caused by our increasingly convoluted building codes have given rise to a growing set of companies that are helping people make sense of the home-building process by incorporating regulations directly into their software.

Using machine learning, their platforms run advanced scenario-analysis around interweaving building codes and inter-dependent structural variables, allowing users to create compliant designs and regulatory-informed decisions without having to ever encounter the regulations themselves.

For example, the prefab housing startup Cover is helping people figure out what kind of backyard homes they can design and build on their properties based on local zoning and permitting regulations.

Some startups are trying to provide similar services to developers of larger scale buildings as well. Just this past week, I covered the seed round for a startup called Cove.Tool, which analyzes local building energy codes – based on location and project-level characteristics specified by the developer – and spits out the most cost-effective and energy-efficient resource mix that can be built to hit local energy requirements.

And startups aren’t just simplifying the regulatory pains of the housing process through building codes. Envelope is helping developers make sense of our equally tortuous zoning codes, while Cover and companies like Camino are helping steer home and business-owners through arduous and analog permitting processes.

Look, I’m not saying codes are bad. In fact, I think building codes are good and necessary – no one wants to live in a home that might cave in on itself the next time it snows. But I still can’t help but ask myself why the hell does it take AI to figure out how to build a house? Why do we have building codes that take a supercomputer to figure out?

Ultimately, it would probably help to have more standardized building codes that we actually clean-up from time-to-time. More regional standardization would greatly reduce the number of conditional branches that exist. And if there was one set of accepted overarching codes that could still set precise requirements for all components of a building, there would still only be one path of regulations to follow, greatly reducing the knowledge and analysis necessary to efficiently build a home.

But housing’s inherent ties to geography make standardization unlikely. Each region has different land conditions, climates, priorities and political motivations that cause governments to want their own set of rules.

Instead, governments seem to be fine with sidestepping the issues caused by hyper-regional building codes and leaving it up to startups to help people wade through the ridiculousness that paves the home-building process, in the same way Concur aids employee with infuriating corporate expensing policies.

For now, we can count on startups that are unlocking value and making housing more accessible, simpler and cheaper just by making the rules easier to understand. And maybe one day my grandkids can tell their friends how their grandpa built his house with his own supercomputer.

And lastly, some reading while in transit:

News Source = techcrunch.com

Artie aims to bring you closer to your digital idols with autonomous AR avatars

in Artificial Intelligence/Augmented Reality/Delhi/India/Politics/TC by

If you spend enough time scrolling through manicured photos of manicured lives on social media, you might come to the realization that maybe the fakeness of the online world has started to look too real.

This might be why so many investors are starting to stare headlong into the world of avatars and digital influencers that aren’t real people but can learn from their audiences in real time. Earlier this week, I chatted with a pair of interesting founders from the startup Artie. The team is basically trying to create an interaction engine for digital avatars to sit in the real world and have some sort of meaningful interaction with users through phone-based AR.

The startup’s backers include Founders Fund and YouTube co-founder Chad Hurley. Co-founders Armando Kirwin and Ryan Horrigan both come from some top startups in the VR media space.

The Artie team

Artie’s sort of autonomous storytelling platform really focuses in on a couple emerging trends.

One is this big idea of digital influencers revving up in Japan and Korea that’s basically leveraging all of these new face-tracking capabilities of smartphones to allow users to craft 3D avatars that are sort of animated, abstracted online personalities. It’s started to make waves stateside, but it’s a slower grind.  Artie isn’t necessarily looking at user-generated content at this moment, but the company’s work in more branded moments with already leveraged IP is an interesting first step towards something bigger.

Artie is also an AR company. The phone AR market really seems to have a number of usage obstacles to overcome. Despite the excitement coming from Apple and Google, platforms like ARKit and ARCore have mostly arrived with a thud. There are a few companies trying to build out some more fundamental backend capabilities to enable shared experiences that adjust to their surroundings, but it’s unclear where the missing link really is in getting people to use a feature that’s really just sitting dormant on their smartphone.

The company is working with WebXR standards that will basically allow anyone to tap a link on their phone and plunge straight into an experience where the avatar is inside their physical space. The video below gives some early insight into what their platform is going to offer.

As niche as this market sounds, Artie isn’t totally alone here, Google has actually flirted with this in its Playground release on Pixel phones where users can jump into photos with 3D characters who are somewhat aware of their environments. For Artie, the deeper interactions between the avatar and characters is really where they hope the magic comes into view. Their platform carries out emotion tracking and object detection to give Unity developers some freedom to let users interrupt the avatars and send them on tangents, all while learning from the user in how they interact with the character and want them to act.

“Think of it like how YouTube, back in the day, established this notion where content creators could for the first time get closer to their audiences through the comments, but it always happens post-mortem after the video was published and would inform what would happen next week,” Horrigan told TechCrunch. “So the difference here is that we’re actually bringing that intimacy between audience and content creator in real time.”

The co-founders both share some big ideas for the direction of storytelling that leverages deep learning to tell the content creators more about the world and audience they’re building for. Artie is at the forefront of some interesting but deeply odd market trends, ones that are probably driven as much by the state of pop culture as they are by tech capabilities, though it’s all still early tech coming from a small team.

The founders say they’ll start working with some early “power users” like media companies and celebrities in the first quarter of next year to start building out the first experiences for Artie on their “Wonderfriend” engine.

News Source = techcrunch.com

The trust dilemma of continuous background checks

in Artificial Intelligence/Asia/background check/Checkr/China/CrunchBase/Delhi/Government/India/intelligo/Masayoshi Son/Meng Wanzhou/Policy/Politics/privacy/Ren Zhengfei/SoftBank/Softbank Vision Fund/Startups/Venture Capital/zte by

First, background checks at startups, then Huawei’s finance chief is arrested, SoftBank’s IPO is subscribed, and I am about to record our next edition of TechCrunch Equity. It’s Thursday, December 6, 2018.

TechCrunch is experimenting with new content forms. This is a rough draft of something new – provide your feedback directly to the author (Danny at danny@techcrunch.com) if you like or hate something here.

The dilemma of continuous background checks

My colleague John Biggs covered the Series A round for Israel-based Intelligo, a startup that provides “Ongoing Monitoring” — essentially a continuous background check that can detect if (when?) an employee has suddenly become a criminal or other deviant. That’s a slight pivot from the company’s previous focus of using AI/ML to conduct background checks more efficiently.

Background checks are a huge business. San Francisco-based Checkr, perhaps the most well-known startup in the space, has raised $149 million according to Crunchbase, driven early on by the need to on-board thousands of contingent workers at companies like Uber. Checkr launched what it calls “Continuous Check” which also actively monitors all employees for potential problems, back in July.

Now consider a piece written a few weeks ago by Olivia Carville at Bloomberg that explored the rise of “algorithmic auditors” that actively monitor employee expenses and flags ones it feels are likely to be fraudulent:

U.S. companies, fearing damage to their reputations, are loath to acknowledge publicly how much money they lose each year on fraudulent expenses. But in a report released in April, the Association of Certified Fraud Examiners said it had analyzed 2,700 fraud cases from January 2016 to October 2017 that resulted in losses of $7 billion.

Here’s a question that bugs me though: we have continuous criminal monitoring and expense monitoring. Most corporations monitor web traffic and email/Slack/communications. Everything we do at work is poked and prodded to make sure it meets “policy.”

And yet, we see vituperative attacks on China’s social credit system, which …. monitors criminal records, looks for financial frauds, and sanctions people based on their scores. How long will we have to wait before employers give us “good employee behavior” scores and attach it to our profiles in Slack?

The conundrum of course is that no startup or company wants (or can) avoid background checks. And it probably makes sense to continually monitor your employees for changes and fraud. If Bob murders someone over the weekend, it’s probably good to know that when you meet Bob at Monday’s standup meeting.

But let’s not pretend that this continuous monitoring isn’t ruinous to something else required from employees: trust. The more heavily monitored every single activity is in the workplace, the more that employees feel that if the system allows them to get away with something, it must be approved. Without any checks, you rely on trust. With hundreds of checks, policy is essentially etched into action — if I can do it, it must meet policy.

In China, where social trust is extremely low, it likely makes sense to have some sort of scoring mechanism to substitute. But for startups and tech companies, building a culture of trust — of doing the right thing even when not monitored — seems crucial to me for success. So before signing up for one of these continuous services, I’d do a double take and consider the potentially deleterious consequences.

If I was a startup employee, I would think twice (maybe thrice?) before traveling to China

Photo by VCG/VCG via Getty Images

Last weekend, Trump and Xi agreed to delay the implementation of tariffs on Chinese goods, which led to buoyant Chinese (tech) stocks Monday in Asia time zones. I wrote about how that doesn’t make any sense, since delaying tariffs doesn’t do anything to solve the structural issues in the US/China conflict:

To me the market is deeply misjudging not only the Chinese economy, but also the American leadership as well.

And specifically, I wrote about constraints on Huawei and ZTE:

In what world do these prohibitions disappear? The U.S. national security agencies aren’t going to allow Huawei and ZTE to deploy their equipment in America. Like ever. Quite frankly, if the choice was getting rid of all of China’s non-tariff barriers and allowing Huawei back into America, I think the U.S. negotiators would walk out.

So it was nice to learn (for me, not for her) that the head of finance of Huawei was arrested last night in Canada at the United States’ request. From my colleague Kate Clark:

Meng Wanzhou, the chief financial officer of Huawei, the world’s largest telecom equipment manufacturer and second-largest smartphone maker, has been arrested in Vancouver, Canada on suspicion she violated U.S. trade sanctions against Iran, as first reported by The Globe and Mail.

Huawei confirmed the news with TechCrunch, adding that Meng, the daughter of Huawei founder Ren Zhengfei, faces unspecified charges in the Eastern District of New York, where she had transferred flights on her way to Canada.

If you wanted to know how the Trump administration was going to continue to fight the trade war outside of tariffs, you now have your answer. This is a bold move by the administration, targeting not just one of China’s most prominent tech companies, but the daughter of the founder of the company to boot.

China has since demanded her return.

Here is how this is going to play out. China is preventing the two American children of Liu Changming from leaving the country, essentially holding them hostage until their father returns to the mainland to face a criminal justice process related to an alleged fraud case. America now has a prominent daughter of a major Chinese company executive in their hands. That’s some nice tit-for-tat.

For startup founders and tech executives migrating between the two countries, I don’t think one has to literally worry about exit visas or extradition.

But, I do think the travel security operations centers at companies that regularly have employees moving between these countries need to keep very keen and cautious eyes on these developments. It’s entirely possible that these one-off “soft hostages” could flare to much higher numbers, making it much more complicated to conduct cross-border work.

Quick Bites

SoftBank’s IPO raises a lot of dollars

KAZUHIRO NOGI/AFP/Getty Images

Takahiko Hyuga at Bloomberg reports that SoftBank has sold its entire book of shares for its whopping $23.5 billion IPO. The shares will officially price on Monday and then will trade on December 19. This is a critical and important win for Masayoshi Son, who needs the IPO of his telecom unit to deleverage some of the risk from SoftBank’s massive debt pile (and also to continue funding his startup dreams through Vision Fund, etc.)

SoftBank Vision Fund math, part 2

Arman and I talked yesterday about the complicated math behind just how many dollars are in SoftBank’s Vision Fund. More details, as Jason Rowley pointed out at Crunchbase News:

In an annual Form D disclosure filed with the Securities and Exchange Commission this morning, SBVF disclosed that it has raised a total of approximately $98.58 billion from 14 investors since the date of first sale on May 20, 2017. The annual filing from last year said there was roughly $93.15 billion raised from 8 investors, meaning that the Vision Fund has raised $5.43 billion in the past year and added six new investors to its limited partner base.

I said yesterday that the fund size should be “$97 billion or $96.7 billion with precision, assuming this $5 billion reaches a final close.” So let’s revise this number again to $99 billion or $98.6 billion with precision, since it seems the $5 billion did indeed close.

What’s next

I am still obsessing about next-gen semiconductors. If you have thoughts there, give me a ring: danny@techcrunch.com.

Thoughts on Articles

Hopefully more reading time tomorrow.

Reading docket

What I’m reading (or at least, trying to read)

  • Huge long list of articles on next-gen semiconductors. More to come shortly.

News Source = techcrunch.com

Where Facebook AI research moves next

in Artificial Intelligence/Delhi/Facebook/India/Politics/TC by

Five years is an awful lot of time in the tech industry. Darling startups find ways to crash and burn. Trends that seem unstoppable sputter-out. In the field of artificial intelligence, the past five years have been nothing short of transformative.

Facebook’s AI Research lab (FAIR) turns five years old this month, and just as the social media giant has left an indelible mark on the broader culture — for better or worse — the work coming out of FAIR has seen some major impact in the AI research community and entrenched itself in the way Facebook operates.

“You wouldn’t be able to run Facebook without deep learning,” Facebook Chief AI Scientist Yann LeCun tells TechCrunch. “It’s very, very deep in every aspect of the operation.”

Reflecting on the formation of his team, LeCun recalls his central task in initially creating the research group was “inventing what it meant to do research at Facebook.”

“Facebook didn’t have any research lab before FAIR, it was the first one, until then the company was very much focused on short-term engineering projects with six-month deadlines, if not less,” he says.

LeCun

Five years after its formation, FAIR’s influence permeates the company. The group has labs in Menlo Park, New York, Paris, Montreal, Tel Aviv, Seattle, Pittsburgh and London. They’ve partnered with academic institutions and published countless papers and studies, many of which the group has enumerated in this handy five-year anniversary timeline here.

“I said ‘No’ to creating a research lab for my first five years at Facebook,” CTO Mike Schroepfer wrote in a Facebook post. “In 2013, it became clear AI would be critical to the long-term future of Facebook. So we had to figure this out.”

The research group’s genesis came shortly after LeCun stopped by Mark Zuckerberg’s house for dinner. “I told [Zuckerberg] how research labs should be organized, particularly the idea of practicing open research.” LeCun said. “What I heard from him, I liked a lot, because he said openness is really in the DNA of the company.”

FAIR has the benefit of longer timelines that allow it to be more focused in maintaining its ethos. There is no “War Room” in the AI labs, and much of the group’s most substantial research ends up as published work that benefits the broader AI community. Nevertheless, in many ways, AI is very much an arms race for Silicon Valley tech companies. The separation between FAIR and Facebook’s Applied Machine Learning (AML) team, which focuses more on imminent product needs, gives the group a “huge, huge amount of leeway to really think about the long term,” LeCun says.

I chatted with LeCun about some of these long-term visions for the company, which evolved into him spitballing about what he’s working on now and where he’d like to see improvements. “First, there’s going to be considerable progress in things that we already have quite a good handle on…”

A big trend for LeCun seems to be FAIR doubling down on work that impacts how people can more seamlessly interact with data systems and get meaningful feedback.

“We’ve had this project that is a question-and-answer system that basically can answer any question if the information is somewhere in Wikipedia. It’s not yet able to answer really complicated questions that require extracting information from multiple Wikipedia articles and cross-referencing them,” LeCun says. “There’s probably some progress there that will make the next generation of virtual assistants and data systems considerably less frustrating to talk to.”

Some of the biggest strides in machine learning over the past five years have taken place in the vision space, where machines are able to parse out what’s happening in an image frame. LeCun predicts greater contextual understanding is on its way.

“You’re going to see systems that can not just recognize the main object in an image but basically will outline every object and give you a textual description of what’s happening in the image, kind of a different, more abstract understanding of what’s happening.”

FAIR has found itself tackling disparate and fundamental problems that have wide impact on how the rest of the company functions, but a lot of these points of progress sit deeper in the five-year timeline.

FAIR has already made some progress in unsupervised learning, and the company has published work on how they are utilizing some of these techniques to translate between languages for which they lack sufficient training data so that, in practical terms, users needing translations from something like Icelandic to Swahili aren’t left out in the cold.

As FAIR looks to its next five years, LeCun contends there are some much bigger challenges looming on the horizon that the AI community is just beginning to grapple with.

“Those are all relatively predictable improvements,” he says. “The big prize we are really after is this idea of self-supervised learning — getting machines to learn more like humans and animals and requiring that they have some sort of common sense.”

News Source = techcrunch.com

1 2 3 84
Go to Top