Connect with us


IBM makes 20 qubit quantum computing machine available as a cloud service

IBM has been offering quantum computing as a cloud service since last year when it came out with a 5 qubit version of the advanced computers. Today, the company announced that it’s releasing 20-qubit quantum computers, quite a leap in just 18 months. A qubit is a single unit of quantum information.

The company also announced that IBM researchers had successfully built a 50 qubit prototype, which is the next milestone for quantum computing, but it’s unclear when we will see this commercially available.

While the earliest versions of IBM’s quantum computers were offered for free to build a community of users, and help educate people on programming and using these machines, today’s announcement is the first commercial offering. It will be available by the end of the year.

Quantum computing is a difficult area of technology to understand. Instead of being built on machines interpreting zeroes and ones in on/off states, quantum computers can live in multiple states. This creates all kinds of new programming possibilities and requires new software and systems to build programs that can work with this way of computing.

Dario Gil, IBM Research VP of AI and IBM Q, says the increased number qubits is only part of the story. The more Qubits you deal with, the more complex the qubit interactions become because they interact with one another in a process called entanglement. If you have more qubits, but there is a high error rate as they interact, then they might not be any more powerful than 5 qubit machine with a lower error rate. He says that IBM researchers have managed to achieve the higher qubit number with low error rates, making them highly useful to researchers. “We have more qubits and less errors, which is combined to solve more problems,” Gil said.

The other issue that comes into play when dealing with quantum states is that they tend to exist for a short period of time in a process known as coherence. It basically means that you only have a brief window of time before the qubits revert to a classical computing state of zeroes and ones. To give you a sense of how this coherence has been progressing, it was just a few nanoseconds when researchers started looking at this in the late 90s. Even as recently as last year, they were able to achieve coherence times of 47 and 50 microseconds for the 5 qubit machines. Today’s quantum machines are in the 90 microsecond range. While that doesn’t sound like much, it’s actually a huge leap forward.

All of these variables make it difficult for a programmer to build a quantum algorithm that can achieve something useful without errors and before it reverts to a classical state, but that doesn’t take away from just how far researchers have come in recent years, and how big today’s announcement is in the quantum computing world.

The ultimate goal of quantum computing is a fault tolerant universal system that automatically fixes errors and has unlimited coherence. “The holy grail is fault-tolerant universal quantum computing. Today, we are creating approximate universal, meaning it can perform arbitrary operations and programs, but it’s approximating so that I have to live with errors and a [limited] window of time to perform the operations,” Gil explained.

He sees this is an incremental process and today’s announcement is a step along along the path, but he believes that even what they can do today is quite powerful. With today’s release and the the improvements that IBM made to the QISKit, a software development kit (SDK) to help companies understand how to program quantum computers, they can continue to advance the technology. It’s not going to happen overnight, but companies, governments, universities and interested parties are undertaking research to see how this can work in practical application. (And of course, IBM isn’t the only company working on this problem.)

IBM sees applications for quantum computing in areas like medicine, drug discovery and materials science as this technology advances and becomes better understood. It is also trying to anticipate possible negative consequences of an advanced technology such as the ability to eventually be able to break encryption. Gil says they are working with standards bodies to try and  develop post-quantum computing encryption algorithms, and while they are a long way from achieving that, they certainly seem to understand the magnitude of the issues and are trying to mitigate them.


Onward helps businesses automate their customer service

How much can customer service be automated? Onward has some straightforward targets — 40 percent of tickets and 40 percent of messages should be automated, and average response times should be 40 seconds on average.

Founders Rémi Cossart and Pramod Thammaiah describe this as Automation40 — basically, a set of goals for businesses looking to bring more automation into the customer service process. They compare these targets to fitness goals: The idea isn’t to hit them right away, but rather to have something to aim for.

This is a new direction (and new name) for Cossart and Thammaiah’s startup. They were previously building Agent Q, a text message-based shopping assistant. Like other founders building virtual assistants, they pitched Agent Q as a mix of automation and human interaction. Eventually, they decided that the real opportunity lay in helping other companies achieve that mix.

To do that, they’ve designed different solutions to address different types of customer service questions.


For the most common questions, Onward can simply pull up a response from the company’s knowledge base. For queries that are a bit more difficult, there’s a visual bot builder, allowing customers to design the flow around how questions get answered, what information gets collected from the customer and so on.

In some cases, the system will need to hand customers over to a human agent. But even in those cases, the agents will be assisted by Onward’s technology, which will suggest different answers, hopefully making them faster and more accurate.

Onward is launching today as a self-serve product. Monthly pricing starts at $9 for the most basic version, going up to $99 per desk for features like integration into HubSpot and Salesforce.
Featured Image: Onward Readmore

Continue Reading


Vectra raises $36M for its AI-based approach to cybersecurity intrusion detection

With the trend of growing cybercrime showing no indication of abating, a startup called Vectra that has built an artificial intelligence-based system called Cognito to detect cyberattacks and mobilise security systems to respond to them has raised $36 million to expand its R&D and business development.

This Series D comes on the back of a strong year for the startup, with 181 percent growth in customer subscriptions between 2016 and 2017, and Vectra’s CEO Hitesh Sheth said he expects the same this year. Typical customers are large enterprises (which is why you don’t see much about pricing on the site) and includes players in the financial, healthcare, government, tech and education sectors. The list the company disclosed to me includes LiveNation/Ticketmaster, Pinterest, Kronos, Tribune Media, Verifone, Agilent, Texas A&M University and DZ Bank in Germany.

This latest round is being led by Atlantic Bridge Capital, with participation from Ireland’s Strategic Investment Fund (ISIF) and Nissho Electronics Corp. Previous investors Khosla Ventures, Accel Partners, IA Ventures, AME Cloud Ventures, DAG Ventures and Wipro Ventures also participated. The company’s total raised to date is $123 million, and while it is not disclosing its valuation, its pre-money valuation of just under $344 million, according to PitchBook, based on its last funding round in March 2016, is likely getting a big boost after the growth it has seen. Also for context, one of its closer competitors, Darktrace, was last valued at $825 million.

Vectra’s growth — and the round that it has raised — underscores one of the bigger challenges in the market at the moment for enterprises and other organizations.

While there are a number of solutions out there for trying to block malicious hackers and their various techniques, and there are systems in place for stopping them when they are found, there is a gap in the market for the moments where cyber criminals evade the best blocks and then proceed to steal data, sometimes for months or more.

The Winter Olympics in Korea, as one recent example, suffered an attack that was only detected after the malicious hackers had already been sucking up data for 120 days.

“One of the issues for enterprises today is that it’s never been more hostile. The operating assumption is that you will get breached,” said Hitesh Sheth, president and CEO of Vectra. His company’s solution, he says, is not to try to change that currently immutable fact, but to drastically shrink the length of an otherwise months-long attack to minutes and hours.  “The only control you really have is what will you do once you are breached.”

Vectra does this using AI. The thinking here is that, if you are working with large enterprises, there are many places, services, apps and end points that need to be assessed for inconsistencies in how they are being queried and used in the network. Systems that are automated and use machine learning to essentially mimic the behavior of security specialists are the best at doing this kind of searching and identification.

Sheth claims that while there are a number of other intrusion and threat detection services out in the market — Darktrace, Cisco’s intrusion detection (built around a number of acquisitions) and RiskIQ being some of them — Vectra is the only one of these that is built on AI algorithms from the ground up. “AI is a bolt-on for most security players, but this is all we do.”

He also says that the other aspect of its service that helps it stand out is its focus on network, rather than end-point, traffic. “If devices are compromised, end point logs are compromised.”

Sheth describes this latest round as its “path to profitability,” where it could be the last one Vectra needs before it tips into the black itself — a big feat for an SaaS service that also has its sights on an IPO longer-term.

“What is a fad in the valley is to raise as much as possible and then some more,” he said. “Investors can win but I’m not sure employees do. You want to rase as much as possible but you need to see how to scale.” He said initially the company wanted to raise between $25 million and $30 million but “interest was super high and it was oversubscribed, so we accommodated investors that we thought would add value.”

The connection with the Irish strategic investment stems out of the fact that Vectra is going to build an R&D center in Dublin. This came first and the investment came second, Sheth said.

The company selected Dublin because it had considered London and Barcelona — there are already three centers in the US, in Austin, Cambridge San Jose — but backed away from the former because of uncertainties around Brexit, and the latter because of political upheaval. Ireland, he believes, will only grow in prominence for its position as the only English-speaking market still fully in the European Union.

“This is an exciting investment for ISIF, which promises significant economic impact for Ireland,” said Fergal McAleavey, head of private equity at ISIF, in a statement. “It is encouraging to see Ireland leverage its emerging expertise in artificial intelligence by attracting businesses such as Vectra that are on the leading edge of technology. With cybersecurity becoming such a critical issue for all organizations, we are confident that Vectra will deliver a strong economic return on our investment while creating high-value R&D employment here in Ireland.”

Meanwhile, company’s growth is what swayed the lead investor.

“We have been impressed by the remarkable growth of Vectra in this fast-moving cybersecurity market,” said Kevin Dillon, managing partner at Atlantic Bridge Capital, in a statement. “The increasing volume, creativity and effectiveness of cyberattacks means that enterprises must adopt AI to automate cybersecurity operations. We look forward to helping the company expand its global enterprise footprint.”
Featured Image: Getty Images Readmore

Continue Reading


Google tries to make Android more enterprise-friendly with new recommendation program

With so many Android devices out there to choose from, it’s not always easy to find one that’s enterprise-friendly. To help alleviate that problem, Google announced the Android Enterprise Recommended program today.

As the name implies, it’s designed to point enterprise IT departments at devices that Google has deemed to be enterprise-ready. This involves a number of criteria, including minimum hardware specifications for Android 7.0 + devices, support for bulk deployment and managed profiles and devices for a consistent application experience across deployed devices. The full list includes all of the minimum app and hardware specifications required to be included in the program.

Photo: Google

The program also requires that within 90 days of Google releasing them manufacturers make security updates available for at least three years. Ninety days feels a bit long if there is a serious vulnerability in the wild, but Google indicated this was not a fixed list and the company would update the requirements as needed over time.

As for the devices they are recommending, these include a broad range of usual suspects, such as BlackBerry KEYone and Motion; Google Pixel, Pixel XL, Pixel 2 and Pixel 2 XL (of course); Huawei Mate 10, Mate 10 Pro, P10, P10 Plus, P10 Lite and P smart; and LG V30 and G6, among others.

Conspicuously missing from this list is anything by Samsung, a company that has programs in place like Knox specifically designed for the enterprise. There are also no HTC phones, but to be fair, the company left the door open for more devices and additional partners to be added over time.

“You can expect more Android Enterprise Recommended devices to be added in the coming weeks and months. Throughout 2018, we will also be applying the Android Enterprise Recommended framework to additional partner types, including OEMs of ‘dedicated’ and rugged devices, mobile carriers, enterprise mobility management (EMM) providers and systems integrators,” Google director of Android Enterprise, David Still wrote in a company blog post announcing the program.

While fewer companies are probably still buying phones for their employees — those kind of programs tended to peak with the old stalwart BlackBerry devices in the days before “Bring Your Own Device” programs popped up — those who do and want to use Android need to have devices that they can manage and deploy in a consistent way. This program is designed to provide a minimum set of specifications to meet that.
Featured Image: nurphoto/Getty Images Readmore

Continue Reading

Member of The Internet Defense League

Subscribe to our Newsletter