Category Archives: Machine learning

Teradata VantageCloud integrated with Microsoft Azure Machine Learning

Teradata has integrated Teradata VantageCloud, an cloud analytics and data platform, with Microsoft Azure Machine Learning (Azure ML). VantageCloud’s scalability, openness and analytics – ClearScape Analytics – combined with Azure ML’s ability to simplify and accelerate the ML lifecycle could help customers unlock the full value of their data, even in the most complex and… Read more »

The post Teradata VantageCloud integrated with Microsoft Azure Machine Learning appeared first on Cloud Computing News.

Image recognition startup joins Google in France

Googlers having funGoogle has continued its charge on the artificial intelligence market through purchasing French image recognition startup Moodstocks, reports

Moodstocks, founded in 2008, develops machine-learning based image recognition technology for smartphones, which has been described by developers as the ‘Shazam for images’. Financials of the agreement have not been confirmed to date.

“Ever since we started Moodstocks, our dream has been to give eyes to machines by turning cameras into smart sensors able to make sense of their surroundings,” Moodstock said on its website. “Today, we’re thrilled to announce that we’ve reached an agreement to join forces with Google in order to deploy our work at scale. We expect the acquisition to be completed in the next few weeks.”

Artificial intelligence is one of the focal points of the Google strategy moving forward, which was confirmed by Google CEO Sundar Pichai during the company’s recent earnings call, though the focus can be dated back to the $625 million DeepMind acquisition in 2014. Although DeepMind is arguably the most advanced AI system in the industry, readers recently confirmed in a poll Google was the leader in the AI segment, it has seemingly been playing catch up with the likes of Watson and AWS whose offerings have been in the public eye for a substantially longer period of time.

The recognition tools are most likely to be incorporated into the Android operating system, though Moodstocks customers will be able to continue to use the service until the end of their subscription. Moodstocks will be incorporated into Google’s R&D centre in France, where the team will work alongside engineers who are focusing on the development of Youtube and Chrome, two offerings where there could be a link to the Moodstocks technology.

“Many Google services use machine learning (or machine learning) to make them simpler and more useful in everyday life – such as Google Translate, Smart Reply Inbox, or the Google app,” said Vincent Simonet, Head of R&D centre of Google’s French unit. “We have made great strides in terms of visual recognition: now you can search in Google Pictures such as ‘party’ or ‘beach’ and the application will offer you good pictures without you and have never needed to categorize them manually.”

Last month, Google also announced it was expanding its machine research team by opening a dedicated office in Zurich. The team will focus on three areas specifically, machine intelligence, natural language processing & understanding, as well as machine perception.

Elsewhere in the industry, Twitter completed the acquisition of Magic Pony last month reportedly for $150 million. Magic Pony, which offers visual processing technology, was one of the more public moves made by the social media network, which could be seen as unusual as the platform lends itself well to the implementation of AI. Microsoft also announced the purchase of Wand Labs, building on the ‘Conversation-as-a-Platform’ proposition put forward by CEO Satya Nadella at Build 2016.

Twitter acquires machine learning start-up Magic Pony

Twitter has stepped up its efforts in the machine learning arena after announcing the acquisition of visual processing technology company Magic Pony.

While the company claims machine learning is central to the brands capabilities, it has been relatively quiet in the market segment in comparison to industry heavy weights such as IBM, Google and Microsoft. This is the third acquisition the team has made in this area, reported to be in the range of $150 million, following the purchase of Whetlab last year and Mad Bits in 2014, compared to Google who acquired Jetpac, Dark Blue Labs and Vision Factory, as well as $500 million on DeepMind, all in 2014.

“Machine learning is increasingly at the core of everything we build at Twitter,” said Jack Dorsey, Twitter CEO. “Magic Pony’s machine learning technology will help us build strength into our deep learning teams with world-class talent, so Twitter can continue to be the best place to see what’s happening and why it matters, first. We value deep learning research to help make our world better, and we will keep doing our part to share our work and learnings with the community.”

The acquisition follows Twitter’s announcement last week advertisers will now be able to utilize emoji keyword targeting for Twitter Ads. Although a simple proposition in the first instance, the new features did open up the opportunity for machine learning enhanced advertising solutions.

Magic Pony, which was founded in 2014 and currently has 11 employees, was acquired to bolster the visual experiences that are delivered across Twitter apps. The team will link up with Twitter Cortex, the in-house machine learning department, to improve image processing expertise.

The technology itself makes use of the abilities of convolutional neural networks to scale-out an image. By taking the information in a picture, the technology imagines a larger and more in-depth image by scaling out the detail which it sees. Much in the same way a human can imagine the rest of a car by seeing the door, the technology learns lessons from previous experiences and applies logical decisions moving forward.

Magic Pony itself was initially supported by investment from Octopus Ventures who have seemingly found a specialty in finding promising AI start-ups. Prior to Magic Pony being acquired by Twitter, Octopus Ventures invested it Evi which was acquired by Amazon in 2012, and SwiftKey which was acquired by Microsoft this year.

“Today marks a great day for the Magic Pony team,” said Luke Hakes, Investment Director at Octopus Ventures. “We’re proud to have believed in the concept early on and to then have had the privilege of joining their journey. The technology Magic Pony has developed is revolutionary and pushes the boundaries of what is possible with AI in the video space.

“The UK continues to grow as the ‘go-to’ place for companies looking to build best in breed AI technology – Octopus has been fortunate to work with the founders of three companies in this space that have gone on to be acquired, with Evi and Amazon, SwiftKey and Microsoft, and now Magic Pony and Twitter. We are excited for the Magic Pony team, but also to take what we have learnt on the last three journeys and help the next generation of entrepreneurs lead the way in the on-going AI revolution.”

Machine learning front and centre of R&D for Microsoft and Google

Dear Future Im Ready, message on paper, smart phone and coffee on tableMicrosoft and Google have announced plans to expand their machine learning capabilities, through acquisition and new research offices respectively, reports

Building on the ‘Conversation-as-a-Platform’ proposition put forward by CEO Satya Nadella at Build 2016, the Microsoft team has announced plans to acquire Wand Labs. The purchase will add weight to the ‘Conversation-as-a-Platform’ strategy, as well as supporting innovation ambitions for Bing intelligence.

“Wand Labs’ technology and talent will strengthen our position in the emerging era of conversational intelligence, where we bring together the power of human language with advanced machine intelligence,” said David Ku, Corporate Vice President of the Information Platform Group on the company’s official blog. “It builds on and extends the power of the Bing, Microsoft Azure, Office 365 and Windows platforms to empower developers everywhere.”

More specifically, Wand Labs adds expertise in semantic ontologies, services mapping, third-party developer integration and conversational interfaces, to the Microsoft engineering team. The ambition of the overarching project is to make the customers experience more seamless by harnessing human language in an artificial environment.

Microsoft’s move into the world of artificial intelligence and machine learning has not been a smooth ride to date, though this has not seemed to hinder investment. Back in March, the company’s AI inspired Twitter account Tay went into melt-down mode, though the team pushed forward, updating its Cortana Intelligence Suite and releasing its Skype Bot Platform. Nadella has repeatedly highlighted artificial intelligence and machine learning is the future for the company, stating at Build 2016:

“As an industry, we are on the cusp of a new frontier that pairs the power of natural human language with advanced machine intelligence. At Microsoft, we call this Conversation-as-a-Platform, and it builds on and extends the power of the Microsoft Azure, Office 365 and Windows platforms to empower developers everywhere.”

Google’s efforts in the machine learning world have also been pushed forward this week, as the team announced dedicated machine learning research based in the Zurich offices, on its blog. The team will focus on three areas specifically, machine intelligence, natural language processing & understanding, as well as machine perception.

Like Microsoft, Google has prioritized artificial intelligence and machine learning, though both companies will be playing catch-up with the likes of IBM and AWS, whose AI propositions have been in the market for some time. Back in April, Google CEO Sundar Pichai said in the company’s earnings call “overall, I do think in the long run, I think we will evolve in computing from a mobile first to an AI first world,” outlining the ambitions of the team.

Google itself already has a number of machine learning capabilities incorporated in its product portfolio, those these could be considered as relatively rudimentary. Translate, Photo Search and SmartReply for Inbox already contains aspects of machine learning, though the team are targeting more complex and accurate competencies.

Elsewhere, Twitter has announced on their blog advertisers will now be able to utilize emoji keyword targeting for Twitter Ads. This new feature uses emoji activity as a signal of a person’s mood or mind set, allowing advertisers to more effectively communicate marketing messages minimizing the potential for backlash of disgruntled twitter users. Although the blog does not state the use of machine learning competencies, it does leave the opportunity for future innovation in the area.

IBM launches weather predictor Deep Thunder for The Weather Company

cloud storm rainIBM’s Weather Company has announced the launch of Deep Thunder to help companies predict the actual impact of various weather conditions.

By combining hyper-local, short-term custom forecasts developed by IBM Research with The Weather Company’s global forecast model the team hope to improve the accuracy of weather forecasting. Deep Thunder will lean on the capabilities of IBM’s machine learning technologies to aggregate a variety of historical data sets and future forecasts to provide fresh new guidance every three hours.

“The Weather Company has relentlessly focused on mapping the atmosphere, while IBM Research has pioneered the development of techniques to capture very small scale features to boost accuracy at the hyper local level for critical decision making,” said Mary Glackin, Head of Science & Forecast Operations for The Weather Company. “The new combined forecasting model we are introducing today will provide an ideal platform to advance our signature services – understanding the impacts of weather and identifying recommended actions for all kinds of businesses and industry applications.”

The platform itself will combine more than 100 terabytes of third-party data daily, as well as data collected from the company’s 195,000 personal weather stations. The offering can be customized to suit the location of various businesses, with IBM execs claiming hyper-local forecasts can be reduced to between a 0.2 to 1.2 mile resolution, while also taking into account other factors for the locality such as vegetation and soil conditions.

Applications for the new proposition can vary from the agriculture to city planning & maintenance to validating insurance claims, however IBM has also stated consumer influences can also be programmed into the platform, meaning retailers could manage their supply chains and understand what should be stocked on shelves with the insight.

Microsoft launches VC to drive inorganic growth

Microsoft To Layoff 18,000Microsoft has announced the launch of Microsoft Ventures, a new capitalist venture arm to engage start-ups and entrepreneurs in areas which the business does not currently operate.

Speaking on the official Microsoft blog, Nagraj Kashyap Corporate VP for the ventures business, highlighted the launch was in line with objectives to identify start-ups which can inspire the next technology evolution, as opposed to supporting the current portfolio and business objectives.

“In Microsoft’s history of engaging with and supporting start-ups, we’ve done a lot of investing, but not a lot of early stage,” said Kashyap. “Because we would often invest alongside commercial deals, we were not a part of the early industry conversations on disruptive technology trends. With a formalized venture fund, Microsoft now has a seat at the table.”

Technology acquisition has become an intense game in recent months, as a host of tech giants have built new business units to identify potential acquisitions. While this might not be considered an unusual business activity, the trends of innovation through acquisition as opposed to organic growth have seemingly becoming more prominent. Earlier this month, HP announced the launch of its own VC business unit, which could be perceived as a means for the business to diversify its portfolio, entering new markets. These new markets could lead to direct competition with HPE.

Microsoft has a history of creating initiatives to aide and invest in start-ups, having launched the Microsoft Accelerator program, which provides tools, technology and consulting, though this unit will aim to sit between the Accelerator and the function which oversees major acquisitions. Initially the team will have a presence in San Francisco New York City and Tel Aviv, and will also look to expand to additional countries in the future.

“Given that the move to the cloud remains the single largest priority for the industry, identifying the bleeding-edge companies who complement and leverage the transition to the cloud is key to our investment thesis,” said Kashyap.

“Companies developing product and services that complement Azure infrastructure, building new business SaaS applications, promoting more personal computing by enriching the Windows and HoloLens ecosystems, new disruptive enterprise, consumer productivity, and communication products around Office 365 are interesting areas from an investment perspective.”

Aside from technologies which can aide the company’s core capabilities, the team will also be responsible for investigating disruptions in more horizontal axis. Security and machine learning were two areas which were identified by Kashyap on the blog. “Our view is outward into the market — we focus on the inorganic growth of Microsoft, looking at where we can provide a step function, versus incremental progress.”

Intel continues to innovate through Itseez acquisition

IntelIntel has continued its strides into the IoT market through the acquisition of Itseez, a computer vision and machine learning company.

Itseez, which was founded by two former Intel employees, specializes in computer vision algorithms and implementations, which can be used for a number of different applications, including autonomous driving, digital security and surveillance, and industrial inspection. The Itseez inclusion bolsters Intel’s capabilities to develop technology which electronically perceive and understand images.

“As the Internet of Things evolves, we see three distinct phases emerging,” said Doug Davis, GM for the Internet of Things Group at Intel. “The first is to make everyday objects smart – this is well underway with everything from smart toothbrushes to smart car seats now available. The second is to connect the unconnected, with new devices connecting to the cloud and enabling new revenue, services and savings. New devices like cars and watches are being designed with connectivity and intelligence built into the device.

“The third is just emerging when devices will require constant connectivity and will need the intelligence to make real-time decisions based on their surroundings. This is the ‘autonomous era’, and machine learning and computer vision will become critical for all kinds of machines – cars among them.”

The acquisition bolsters Intel’s capabilities in the potentially lucrative IoT segment, as the company continues its efforts to diversify its reach and enter into new growth markets. Last month, CEO Brian Krzanich outlined the organizations new strategy which is split into five sections; cloud technology, IoT, memory and programmable solutions, 5G and developing new technologies under the concept of Moore’s law. Efforts have focused around changing the perception of Intel from a PCs and mobile devices brand, to one which is built on a foundation of emerging technologies.

Intel’s move would appear to have made the decision of innovation through acquisition is a safer bet than organic, in-house innovation. There have been a small number of examples of organic diversification; Apple’s iPhone is one example, though the safer bet to move away from core competence is through acquisition.

Intel has dipped its toe into organic diversification, as it attempted to develop a portfolio of chips for mobile devices, though this would generally not be considered a successful venture, similar to Google’s continued efforts to organically grow into social, which could be seen as stuttering. On the contrary, Google’s advertising revenues now account for $67.39 billion (2015), with its platform being built almost entirely on acquisitions. The AdSense and Adwords services have been built and bolstered through various purchases including Applied Semantics ($102 million in 2003), dMarc Broadcasting ($102 million in 2006), DoubleClick ($3.1 billion in 2007), AdMob ($750 million in 2009) and Admeld ($400 million in 2011).

While diversification through acquisition can be seen as the safer, more practical and efficient means to move into new markets, it is by no means a guaranteed strategy. Intel’s strategy could be seen as a sensible option as there are far more examples off successful diversification through acquisition compared to organic growth. The jury is still out on Intel’s position in the IoT market but there are backing the tried and tested route to diversification.

Pfizer utilizes IBM Watson for Parkinson’s research

healthcare ITIBM and Pfizer have announced a research collaboration with the intention of improving how clinicians deliver care to Parkinson’s patients.

The collaboration will be built on a system of sensors, mobile devices, and IBM Watson’s machine learning capabilities, to provide real-time disease symptom information to clinicians and researchers. The team aim to gain a better understanding as to how the disease progresses as well as how patients react to certain medications, to design future clinical trials and also speed up the development of new therapies.

“We have an opportunity to potentially redefine how we think about patient outcomes and 24/7 monitoring, by combining Pfizer’s scientific, medical and regulatory expertise with IBM’s ability to integrate and interpret complex data in innovative ways,” said Mikael Dolsten, President of Pfizer Worldwide R&D.

According to the World Health Organization, neurological diseases such as Parkinson’s affect almost one billion families around the world, Approximately 60,000 Americans are diagnosed with Parkinson’s disease each year according to the Parkinson’s Disease Foundation, and an estimated seven to 10 million people suffer from the disease globally.

“The key to our success will be to deliver a reliable, scalable system of measurement and analysis that would help inform our clinical programs across important areas of unmet medical need, potentially accelerating the drug development and regulatory approval processes and helping us to get better therapies to patients, faster,” said Dolsten.

The collaboration seeks to create a holistic view of a patient’s well-being by seeking to accurately measure a variety of health indicators. Data generated through the system could also arm researchers with the insights and real-world evidence needed to help accelerate potential new and better therapies.

“With the proliferation of digital health information, one area that remains elusive is the collection of real-time physiological data to support disease management,” said Arvind Krishna, SVP at IBM Research. “We are testing ways to create a system that passively collects data with little to no burden on the patient, and to provide doctors and researchers with objective, real-time insights that we believe could fundamentally change the way patients are monitored and treated.”

Google plays catch-up with Cloud Machine Learning

AI-Artificial-Intelligence-Machine-Learning-Cognitive-ComputingGoogle has entered into the machine learning market with the alpha release of Cloud Machine Learning.

Built on top of the company’s open source machine learning system TensorFlow, the offering will allow customers to build custom algorithms the make predictions for their business, aiding decision making.

“At Google, researchers collaborate closely with product teams, applying the latest advances in machine learning to existing products and services – such as speech recognition in the Google app, search in Google Photos and the Smart Reply feature in Inbox by Gmail,” said Slaven Bilac, Software Engineer at Google Research. “At GCP NEXT 2016, we announced the alpha release of Cloud Machine Learning, a framework for building and training custom models to be used in intelligent applications.”

The system is already used in a number of Google’s current offerings, though it is later to market than its competitors. AWS launched its machine learning in April last year, while IBM’s Watson has been making noise in the industry for years.

Although later to market, Google has highlighted that it will allow customers to export their TensorFlow models to use in other settings, including their own on premise data centres. Other offerings operate in vendor lock-in situation, meaning their customers have to operate the machine-learning models they’ve built in the cloud through an API. Industry insiders have told BCN that avoiding vendor lock-in situations would be seen as a priority within their organization, which could provide Google with an edge in the machine-learning market segment.

Cloud Machine Learning’s launch builds on the growing trend towards advanced data analytics and the use of data to refine automated decision making capabilities. A recent survey from Cloud World Forum showed that 85% of respondents believe data analytics is the biggest game changer for marketing campaigns in the last five years, while 82% said that data would define the way in which they interact with customers.

The company is still behind Microsoft and AWS in the public cloud space, though recent moves are showing Google’s intent to close the gap. At GCP NEXT 2016, Google’s cloud chief Diane Greene told the audience that machine learning and security will form the back bone of her new sales strategy. “If your customer is embracing machine learning, it’d be prudent for you to embrace it too,” said Greene.

HPE launches ‘machine-learning-as-a-service’ on Microsoft Azure

HPE office logoHPE has upgraded its Haven OnDemand proposition to deliver it as ‘machine learning as a service’ via Microsoft Azure.

The product offers a freemium model and has collected around 12,000 registered developers since the beta launch in 2014. Through the leadership of Haven OnDemand CTO, Chris Goodfellow, the service is built on the mantra of ‘the sum is greater than the parts’, utilizing more than 60 API’s which combine to provide machine learning capabilities.

“The software industry is on the cusp of a new era of breakthroughs, driven by machine learning that will power data-driven applications across all facets of life,” said Colin Mahony, GM of HPE Big Data. “HPE Haven OnDemand democratizes big data by bringing the power of machine learning, traditionally reserved for high-end, highly trained data scientists, to the mainstream developer community”

Haven OnDemand includes features designed for applications such as sentiment analysis in text, text extraction from images, face and logo recognition, social media analysis and speech recognition. Developers can also build a set of self-learning functions that analyze, predict and alert based on structured datasets. French start-up Ayni utilized the speech recognition API to help it create text transcripts of live audio streams on its foreign language education app.

Alongside the product development over the last 12 months, HPE has also run an active global hackathon program, which has provided feedback to help optimize the offering.

All HPE Haven OnDemand APIs and services are hosted on Microsoft Azure, building on the long-term strategic partnership between the two tech giants. Back in December, the partnership was extended as HPE appointed Microsoft Azure as a preferred public cloud partner. In return, HPE was granted preferred partner status in providing infrastructure and services for Microsoft hybrid cloud offerings.

“Organizations have massive quantities of information that can hold insights into business transformation, but harnessing it can be challenging,” said Garth Fort, General Manager, Partner and Channel Marketing, Cloud and Enterprise at Microsoft. “Leveraging the high performance and scalability of Azure, HPE Haven OnDemand brings our mutual customers a compelling solution to help turn their data into value.”