Powered by Blogger.
RSS

All 8 of the Microsoft HoloLens enterprise customer use cases ranked from best to worst

Microsoft has launched pre-orders for UK developers to get their hands on their latest piece of hardware, the mixed reality HoloLens headset.

HoloLens is not a consumer VR headset. For starters it costs £2,700, so the current use cases all apply to industry verticals, be it medicine, manufacturing, architecture, travel and leisure or entertainment.

The problem, and this is with HoloLens in general, is the virtual field of vision is so small, roughly like looking at a 24 inch screen from a metre away, that you can't immerse yourself in the surroundings, making the use cases very object specific, such as a jet engine or the human body.

Here we have ranked the eight HoloLens partner case studies listed on the Microsoft website so far, in a completely arbitrary way, from enterprise ready to shameless marketing gimmick.

1. Japan Airlines

Japan Airlines
© YouTube/Microsoft

Japan Airlines is using HoloLens to allow trainee engineers to view working holograms of jet engines without having to visit a hangar or take an expensive working component out of service.

Lauren Bissett from the HoloLens team starts her demo by saying "employees can experience realistic 3D training at scale. Learning to work on complex machinery that is hard to access, like a jet engine". Trainees can collaborate on the hologram either in the classroom or via Skype.

Verdict: This is pretty useful, and the point about training on components that aren't easy to get your hands on is very valid. A global company that uses Skype for training and teacher/student communications could definitely benefit from this sort of set up.

2. NASA

NASA
© YouTube/Microsoft

NASA's jet propulsion laboratory collaborated with Microsoft to create a piece of software called OnSight. As detailed by Jeff Norris, the OnSight project lead, they use HoloLens "to connect scientists and engineers with the environment of the Curiosity Mars Rover. Because we can't put our scientists physically on Mars, technology like this allows us to investigate what is possible if we can make them virtually present".

What this means is allowing staff to view a 3D image of the mapped surface of Mars and, "the plan is to deploy on site to mission operations this summer to be controlling Rovers on Mars in July", says Norris.

Verdict: This is the peak use case Microsoft will want to roll out, because it is so unashamedly nerdy. It's difficult to know how immersive the Mars environment is in HoloLens without trying it but my concern would be the limitations imposed by the small viewing window, as opposed to a fully immersive VR experience.

What is interesting here is NASA saying that it will use the tool to actually control the Rover, which would require some pretty comprehensive retooling but would provide one of the more practical use cases the HoloLens has so far in its limited lifetime.

3. Volvo

Volvo
© YouTube/Microsoft

Swedish car makers Volvo are using HoloLens in a couple of ways, allowing customers to view modifications on a 3D model of the car as well as for internal training purposes.

Verdict: Having a HoloLens at the dealership sounds like a perfect use case for the technology. You only need one unit and you can show customers all of their modifications on a scale model, which might drive purchases or at least will drive the perception that Volvo is an innovative car brand.

The video didn't go into detail about the training, design or maintenance use cases in much detail but there is certainly some scope for the technology in all of those areas.

4. Trimble

Trimble
© YouTube/Microsoft

Aviad Almagor, MR program manager at GPS technology specialists Trimble opens the video by saying: "Architects are dealing with shapes, and spaces and light and they dream in 3D and need to translate that into 2D documents. HoloLens presents a whole new paradigm."

The core to this use case is human's apparent inability to visualise in 3D. What Trimble does with HoloLens is allow architects and structural engineers to view holographic models of structures, be it layered on top of a physical model or actually on site to aid decision making.

Almagor says on the video that HoloLens is a "much better way to make sure the design is implemented correctly on site".

Verdict: This is a pretty solid use case and I see it being more popular in the setting of an architect office, stood around a 3D model, than I do on the site itself.

5. Case Western University

Case Western University
© YouTube/Microsoft

Case Western Reserve University in Cleveland, Ohio is trialling lectures using HoloLens so that students can "stand around a model like a tour group in a museum", according to Mark Griswold, a professor of radiology at the university.

The application allows students to view a 3D hologram of the human body and cycle through different layers of anatomical structures, from muscular to circulatory systems and bone structure.

Verdict: It definitely sounds good, but in practice (I have tried it) the 3D model is pretty underwhelming. The detail isn't quite there yet and feels more appropriate for a GCSE biology classroom than a medical training university. My guess is that students using HoloLens will need to get their hands on a cadaver at some point also, making the whole exercise an expensive extra step.

6. Autodesk

Autodesk
© YouTube/Microsoft

Design software maker Autodesk is a natural fit for the HoloLens. On the video it says: "The number one challenge for designers and engineers is just being on the same page which is key to making better decisions."

It goes on to show two designers drag a 3D model of a game console into a space between their desk to collaborate on the design. It has also developed an application called ModBot to visualise industrial scale robots to make better decisions pre-build and deployment.

Verdict: The part where two designers drag a 3D model of a game controller into a space between their desks is pretty cool. My question is, why can't they just collaborate on a real piece of kit? Wouldn't that be a better use of time and resources?

7. PGA Tour

PGA Tour
© YouTube/Microsoft

Microsoft presented a sporting use case at its Worldwide Partner Conference earlier this year with the PGA Tour.

In partnership with digital studio TaqTile they created an app for fans and players to view a 3D rendering of golf courses. In the video the presenter shows a 16ft hologram of TPC Sawgrass rendered from 3D Bing maps. It could be navigated by the air pinch or by voice, "go to the 17th hole" and individual player shot arcs and locations could be layered on top.

Scott Gutterman, VP of digital operations at the PGA Tour said they are looking at using the HoloLens for golf course design and tournament set up, so visualising grandstand locations and traffic flows to traverse the course. Players and caddies could also use the technology to strategise how they will attack the course.

Verdict: As a golfer this looks really fun to play with but it is essentially an advanced sports analytics application which would require me to put a HoloLens on while at a major golf event, so it is pretty restrictive.

8. Legendary Entertainment

Legendary Entertainment
© YouTube/Microsoft

The makers of such masterpieces as the Warcraft movie, media company Legendary Entertainment has designed a slick HoloLens use case video which shows employees sat at desks interacting with World of Warcraft character holograms.

"Legendary, although primarily known as a film and television company we really see ourselves as world builders in the experience business," Barnaby Legg, VP of theatrical strategy says in the video.

The video also shows pre-visualisations of sets and characters and some cool social media ready 'content' for fans.

Verdict: The stuff about artists pre-visualising sets and characters is pretty cool, and potentially useful for people in the creative industries.

What this use case boils down to though is having attendees of red carpet events put a HoloLens on to 'meet' characters from the movie, which Legg calls a "magic trick" during the video.

  • Digg
  • Del.icio.us
  • StumbleUpon
  • Reddit
  • RSS

16 innovative businesses using IBM Watson: Which companies are using Watson's big data and analytics to power their business?

Watson is the supercomputer that famously beat human contestants in the US TV show Jeopardy five years ago. Initially a research project, the cognitive __computer is being used across industries the world over to boost revenue, efficiency and even save lives.

IBM Watson provides cloud based predictive analytics for business insight, and with the cognitive computing services embedded in Watson's analytics engine, users can ask natural language questions and visualise their data patterns. 

Here are the top 16 ways businesses are using IBM Watson...

1. IBM Watson: General Motors

IBM Watson: General Motors
Credit: iStock/wendellandcarolyn

In October 2016, IBM and General Motors (GM) joined forces to add artificial intelligence to GM's cars. The partnership aims to offer location-based products and services to you while you're in your car.

GM's OnStar Go, is the industry's first cognitive mobility service and will use machine learning to understand user preferences, and recognise patterns found in your decision data. From that, customers will receive personalised marketing services from numerous partners such as Glympse, iHeartRadio, Mastercard and Parkopedia.

2. IBM Watson: Harrow Council

IBM Watson: Harrow Council
Credit: iStock/Memitina

In September 2016, IBM Watson signed a 10-year contract with Harrow Council to bring Watson Care Manager to support individuals in the UK.

IBM's Watson Care Manager can control scheduling, develop personalised care plans, manage budgets, select providers and enable care. This will allow individuals and caregivers to quickly choose the most appropriate healthcare provider, determined by their allocated personal budget.

According to IBM, this is the first implementation of Watson Care Manager outside of the US.

3. IBM Watson: Condé Nast

IBM Watson: Condé Nast
Credit: iStock/David_Ahn

In September 2016, Condé Nast partnered with IBM Watson to help build informed social media campaigns for its brands. The software built by IBM and Influential, a 'data-first influencer platform' offers Condé Nast customers (such as the New Yorker and Vogue) insight into who to target their campaigns towards and what celebrities would make good brand ambassadors. 

For example, if a brand wants to find somebody who is compassionate, Watson will analyse the last 20,000 words and emojis potential 'influencers' have published looking for the perfect person fitting the breif. 

4. IBM Watson: GSK

IBM Watson: GSK
Credit: GSK

Pharmaceutical company GlaxoSmithKline (GSK) is teaming up with IBM Watson to better connect with its customers, enabling people to ask questions by voice and text right via GSK's online ads.

"Watson provides a very personalised experience" senior brand manager of the Cough and Cold division at GSK, Jason Andree told Computerworld earlier this year. "If you're sick, through Watson, you can ask a question and it will provide a personalized response."

5. IBM Watson: Macy’s On-Call

IBM Watson: Macy’s On-Call

In July 2016, US retailer Macy's announced it's testing 'Macy's On-Call', a new mobile service in which shoppers can ask (in natural language) Watson questions about a store's products, services and facilities. 

This technology is delivered by Satisfi, a  location-based engagement software which accesses Watson from the cloud, providing shop assistance in both English and Spanish.

Macy’s chief growth officer, Peter Sachse said: "This program, in partnership with Satisfi and leveraging the power of IBM Watson, will help us explore new ways to engage one-on-one with customers in-store, providing them another level of service right at their fingertips."

6. IBM Watson: American Cancer Society

IBM Watson: American Cancer Society

In April 2016, IBM Watson partnered with the American Cancer Society to create a virtual adviser that uses machine learning to offer patients personalised information and advice.

To help determine the relevant advice to give, the advisor will look at the patient's type of cancer, its stage and their previous treatments enabling patients to ask questions in natural language and receive audio responses. 

IBM Watson will use data from Cancer.org and the cancer society's National Cancer Information centre to provide meaningful answers to patient's questions. The advisors will also take data from IBM's Watson Health Cloud.

American Cancer Society CEO, Gary M. Reedy said: "This partnership can take these efforts to next level by combining the depth and breadth of cancer information from the world's most trusted cancer source with the power of cognitive technology from IBM Watson."

7. IBM Watson: Nonprofit educational organisation Sesame Workshop

IBM Watson: Nonprofit educational organisation Sesame Workshop
© iStock/Beyhan Yazar

IBM Watson and Sesame Workshop announced a partnership to improve early childhood education around the globe. As part of a three-year agreement, this collaboration will see IBM and Sesame Workshop create new educational products and systems that can adapt to the learning styles or children and change according to each individual child's aptitude level. 

IBM’s general manager for Watson IoT, Harriet Green said: "The potential for Watson to absorb, correlate, and learn from huge amounts of unstructured data and then deliver very personalized educational experiences is unprecedented. Working together with Sesame Workshop, we aim to transform the way in which children learn and teachers teach, and envision having an impact on the lives and education of millions of children."

More information here.

8. IBM Watson: Imperial College London predicts crime with Watson

IBM Watson: Imperial College London predicts crime with Watson

Back in 2014 University students at Imperial College London were able to use supercomputer Watson to analyse and predict crime.

Computing students in Imperial College London can work with IBM and team Watson to develop a cognitive __computer application to solve challenging problems within the university’s crime department.

9. IBM Watson: Mears Group

IBM Watson: Mears Group
© iStock/Azmanjaka

UK housing and social care provider, Mears Group has over 700,000 social homes to maintain across the UK and 20,000 employees. The Mears Group also has over 600 million health and safety/accident reports on their database.

Last year Mears Group integrated with IBM Watson to gain faster reporting turnaround and new insights and revenue opportunity suggestions.

10. IBM Watson: University of Southampton

IBM Watson: University of Southampton

The University of Southampton is one of the first European university to collaborate with IBM Watson to offer research modules for students from all disciplines like medicine, chemistry, marine studies and the arts, rather than the typical computing subjects.

Southampton students get to use Watson in a new Cognitive Computing module as part of its Curriculum Innovation programme which will initially be available to over 150 undergraduate and Master’s students across all disciplines. 

11. IBM Watson: Citigroup

IBM Watson: Citigroup
© iStock/Ymgerman

In 2012 the US based multinational banking company Citigroup explored Watson's analytics with the aim to improve customer relationships and interactions in the bank. 

Citigroup aimed to use Watson to gather vast information from customers and provide suggestions to improve customer interactions, evaluated risk and identify opportunities and data patterns.

12. IBM Watson: Cloud start-ups get free credits and access to Watson APIs

IBM Watson: Cloud start-ups get free credits and access to Watson APIs

IBM launched its global entrepreneur program for cloud startups in 2014. It offers startups $120,000 worth of free cloud credits, as well as connections into IBM’s vast global network of enterprise clients, consultants, innovation centres and more.

IBM unleashed its Watson APIs on Bluemix, IBM's open, public cloud development platform - giving any developer with an internet connection the power to build with the latest cognitive computing technology, spurring them to develop apps which can think, learn and solve problems. 

13. IBM Watson: Science and Technology Facilities Council uses Watson to help GlaxoSmithKline, emergency services and construction business

IBM Watson: Science and Technology Facilities Council uses Watson to help GlaxoSmithKline, emergency services and construction business

Science & Technology Facilities Council’s (STFC) Hartree Centre has created a big data innovation hub using IBM Big Data and Analytics technology to help organisations solve complex business problems and achieve competitive advantage.

Hartree Centre provides advanced computational and data analytics facilities as well as access to data scientists for customers looking to tackle masses of unstructured data.

Democrata, a UK construction company – is using the powerful analytics at Hartree Centre to automate environmental impact assessments and better predict risk. Surveys are required to analyse a multitude of data sets to understand the impact of a high speed rail link or a new road. In the past this has been a slow and cumbersome task. But now, using big data analytics automated searches and queries on disparate data sets means faster and cheaper decision making for large engineering and construction firms. 
  
KnowNow Information, an IBM Business partner is helping local government emergency service providers improve their decision making and resources planning. Using the Hartree Centre big data analytics technology and data scientists, the emergency services provider can now see patterns of incidents across their territory and the system can begin to predict where and when to allocate resources. This has the potential to save time and money for all emergency response teams.   
  
Global healthcare company GSK is working with the Hartree Centre to develop clustering techniques similar to that used on social networking sites to enhance the understanding of these relationships. ‘Disease mapping’ looks at a wide array of source data from various journals and other records to identify correlations between genes, biological processes and known diseases. 

14. IBM Watson: Cleveland Clinic

IBM Watson: Cleveland Clinic

US-based not-for-profit the Cleveland Clinic has begun a pilot program using IBM's Watson cognitive computing technology (through the cloud) to research new cancer treatments.

Typically, doctors do not have the time or resources to research specific treatments based on a patients cancer type along with their DNA. The Cleveland Clinic hopes to change this, and IBM's Chairman and CEO Ginni Rometty said researchers at the clinic are using IBM's Watson Genomics Analytics to track patient's genes.

15. IBM Watson: Standard Bank

IBM Watson: Standard Bank

Standard Bank uses IBM Watson to speed handling of customer queries, allowing it to identify customers quickly so they can respond in faster time. 

"The ultimate beneficiaries of the project will be our customers for whom the process-known as 'cognitive computing' will undoubtedly bring many benefits as we continue to identify innovative ways of doing business and build a bank for the future,” said Vuyo Mpako, head, innovation and channel design, Standard Bank.

16. IBM Watson: Under Armour

IBM Watson: Under Armour
© Pixabay

Sportswear brand Under Armour has partnered with IBM to produce in-depth health and fitness insights though digital health and fitness applications and powered by Watson. 

Under Armour has created an app using Watson's analytics to pool personal, physiological, behavioural and nutritional data. The app will act as a personal trainer, providing users with suggestions for optimal health and fitness based on Watson's analytic insights.

  • Digg
  • Del.icio.us
  • StumbleUpon
  • Reddit
  • RSS

Making computers explain themselves

  • Researchers from MIT’s   Science and Artificial Intelligence Laboratory (CSAIL) have devised a way to train neural networks so that they provide not only predictions and classifications but rationales for their decisions.

    Researchers from MIT’s __computer Science and Artificial Intelligence Laboratory (CSAIL) have devised a way to train neural networks so that they provide not only predictions and classifications but rationales for their decisions.

    Illustration: Christine Daniloff/MIT

    Full Screen
  • In recent years, the best-performing systems in artificial-intelligence research have come courtesy of neural networks, which look for patterns in training data that yield useful predictions or classifications. A neural net might, for instance, be trained to recognize certain objects in digital images or to infer the topics of texts.

    But neural nets are black boxes. After training, a network may be very good at classifying data, but even its creators will have no idea why. With visual data, it’s sometimes possible to automate experiments that determine which visual features a neural net is responding to. But text-processing systems tend to be more opaque.

    At the Association for Computational Linguistics’ Conference on Empirical Methods in Natural Language Processing, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) will present a new way to train neural networks so that they provide not only predictions and classifications but rationales for their decisions.

    “In real-world applications, sometimes people really want to know why the model makes the predictions it does,” says Tao Lei, an MIT graduate student in electrical engineering and __computer science and first author on the new paper. “One major reason that doctors don’t trust machine-learning methods is that there’s no evidence.”

    “It’s not only the medical domain,” adds Regina Barzilay, the Delta Electronics Professor of Electrical Engineering and Computer Science and Lei’s thesis advisor. “It’s in any domain where the cost of making the wrong prediction is very high. You need to justify why you did it.”

    “There’s a broader aspect to this work, as well,” says Tommi Jaakkola, an MIT professor of electrical engineering and computer science and the third coauthor on the paper. “You may not want to just verify that the model is making the prediction in the right way; you might also want to exert some influence in terms of the types of predictions that it should make. How does a layperson communicate with a complex model that’s trained with algorithms that they know nothing about? They might be able to tell you about the rationale for a particular prediction. In that sense it opens up a different way of communicating with the model.”

    Virtual brains

    Neural networks are so called because they mimic — approximately — the structure of the brain. They are composed of a large number of processing nodes that, like individual neurons, are capable of only very simple computations but are connected to each other in dense networks.

    In a process referred to as “deep learning,” training data is fed to a network’s input nodes, which modify it and feed it to other nodes, which modify it and feed it to still other nodes, and so on. The values stored in the network’s output nodes are then correlated with the classification category that the network is trying to learn — such as the objects in an image, or the topic of an essay.

    Over the course of the network’s training, the operations performed by the individual nodes are continuously modified to yield consistently good results across the whole set of training examples. By the end of the process, the computer scientists who programmed the network often have no idea what the nodes’ settings are. Even if they do, it can be very hard to translate that low-level information back into an intelligible description of the system’s decision-making process.

    In the new paper, Lei, Barzilay, and Jaakkola specifically address neural nets trained on textual data. To enable interpretation of a neural net’s decisions, the CSAIL researchers divide the net into two modules. The first module extracts segments of text from the training data, and the segments are scored according to their length and their coherence: The shorter the segment, and the more of it that is drawn from strings of consecutive words, the higher its score.

    The segments selected by the first module are then passed to the second module, which performs the prediction or classification task. The modules are trained together, and the goal of training is to maximize both the score of the extracted segments and the accuracy of prediction or classification.

    One of the data sets on which the researchers tested their system is a group of reviews from a website where users evaluate different beers. The data set includes the raw text of the reviews and the corresponding ratings, using a five-star system, on each of three attributes: aroma, palate, and appearance.

    What makes the data attractive to natural-language-processing researchers is that it’s also been annotated by hand, to indicate which sentences in the reviews correspond to which scores. For example, a review might consist of eight or nine sentences, and the annotator might have highlighted those that refer to the beer’s “tan-colored head about half an inch thick,” “signature Guinness smells,” and “lack of carbonation.” Each sentence is correlated with a different attribute rating.

    Validation

    As such, the data set provides an excellent test of the CSAIL researchers’ system. If the first module has extracted those three phrases, and the second module has correlated them with the correct ratings, then the system has identified the same basis for judgment that the human annotator did.

    In experiments, the system’s agreement with the human annotations was 96 percent and 95 percent, respectively, for ratings of appearance and aroma, and 80 percent for the more nebulous concept of palate.

    In the paper, the researchers also report testing their system on a database of free-form technical questions and answers, where the task is to determine whether a given question has been answered previously.

    In unpublished work, they’ve applied it to thousands of pathology reports on breast biopsies, where it has learned to extract text explaining the bases for the pathologists’ diagnoses. They’re even using it to analyze mammograms, where the first module extracts sections of images rather than segments of text.

    “There’s a lot of hype now — and rightly so — around deep learning, and specifically deep learning for natural-language processing,” says Byron Wallace, an assistant professor of computer and information science at Northeastern University. “But a big drawback for these models is that they’re often black boxes. Having a model that not only makes very accurate predictions but can also tell you why it’s making those predictions is a really important aim.”

    “As it happens, we have a paper that’s similar in spirit being presented at the same conference,” Wallace adds. “I didn’t know at the time that Regina was working on this, and I actually think hers is better. In our approach, during the training process, while someone is telling us, for example, that a movie review is very positive, we assume that they’ll mark a sentence that gives you the rationale. In this way we train the deep-learning model to extract these rationales. But they don’t make this assumption, so their model works without using direct annotations with rationales, which is a very nice property.”


    • Digg
    • Del.icio.us
    • StumbleUpon
    • Reddit
    • RSS

    IBM offers dos and don'ts for service providers winning new OpenStack customers

    Speaking at the OpenStack Summit in Barcelona, John Arwe, a member of IBM’s senior technical staff, offered up some pointers on the do’s and do not do’s of working with customers who are interested in OpenStack. Here we present a few of them – with some small edits from Computerworld UK for conciseness.

    Arwe offers some insights from the service provider's point of view that will be useful for IT teams and CIOs to consider as well.

    Do: Define the cloud

    "One of the first problems we had was our own clients were being told by their CIOs you have got to be in the cloud. And sometimes the CIO kind of knows what that means, and just as often not really, let alone if their definition of the cloud is the same as the OpenStack definition of the cloud," Arwe said.

    "Our own technical leaders in some cases, sales people in others, some of them had been doing virtualisation since the 1970s and they have their homegrown processes. Some people would say sure you can do cloud, you don’t have to change any of that other stuff – don’t worry about it.

    "If you know OpenStack you know that’s not really true, OpenStack has its ideas: this is my playground, everybody else get off. So you have to be careful about that. Probe them a bit and see what they mean – if they say we want to do cloud, find out what that means to them."

    Don’t: Forget your clients might not have a technical background

    "The questions one client didn’t know they needed to ask was: are you willing to give up custom scripts and things that you used to build and manage your guests for years, that are hooked up into your compliance systems?

    "The answer of course was no, we don’t want to give those things up, so there’s an organisational debate going on. You have to figure out how you want to do compliance – you don’t have to do it on OpenStack. You can’t expect the kinds of enterprise clients you’re going to get, especially from executives or managers of IT infrastructure groups, to understand all the technical details of OpenStack."

    Do: Distinguish between public and private clouds

    "The third thing is the assumption public and private clouds have the same needs. Another way to think about this is: are you trying to be a service provider or just try to run your own devops stuff in the cloud? They are very different business models – and that forces different effects on you. There are very different usage models and you have to have different metrics sometimes for your own development group."

    Do: Find out what your customer needs, specifically

    "Another thing that can go wrong is why are they doing this – what do they really need out of it? One of the criteria that blindsided some people in a development organisation is they came in and said: 'Alright, we have been working on this for two months, it’s just about there, we’re ready to demo to the executives'. And this other group demoed with another cloud provider and they were able to provision 20 in 10 minutes – the moral is, you have to understand what your competition is, even within the company."

    Do: Demonstrate how you’ll make life easier for end users, as well as operators

    "People resist change naturally. If you can provide something to them in a way that makes lives easier, your conversations are going to be so much better.

    "It’s not just using OpenStack with the people running it, the operators – it’s the end users. Keep the end users happy, and the operators are happy, their managers are happy, and they’ll buy your stuff, oddly enough. So you have to keep in mind that fact, not just making the operators happy but also the people the operators are dealing with."

    Don’t: Underestimate the importance of regional regulations

    "Your clients have to deal with compliance – in the US there are healthcare rules, in Europe there’s rules about where your data is – all these things matter more or less [depending on] which government you happen to be dealing with at the time or which industry.

    "If it takes them three or four months to get through their internal certification process to make a new VM go online – which is not uncommon in the enterprise space – and you tell them every six months you’re going to be doing a complete upgrade of the whole thing, they kind of toss a hairball on that."

    • Digg
    • Del.icio.us
    • StumbleUpon
    • Reddit
    • RSS

    IT salary expectations for 2017 UK - how much should UK IT pros be earning next year?

    How much should you earn next year?  Read on to see how salaries are set to increase for IT professional jobs in the UK - whether you are a CIO, network engineer or mobile developer...

    IT professionals' salaries are set to rise next year as UK firms invest in growing their talent pool, promoting digitisation across whole organisations. This will result in higher demand for IT pros with both technical and organisational skills.

    That is according to recruitment specialist Robert Half's latest 2017 Salary Guide, which claims that salaries for top IT talent will continue to rise in 2017. The report states that 90 percent of IT directors struggle to find skilled professionals today, and are willing to pay a higher salary for those with the right expertise.

    Next year should see growing demand for security engineers, front-end developers and networking engineers, with cloud technology and IT security remaining a priority for CIOs and senior IT leaders.

    The report also claims that demand for agile development, support and maintenance will result in continued growth in devops in 2017 as 'businesses look for a leaner way of supporting applications'.

    So which roles will see the biggest increase? Read on to see what you should be paid in 2017…

    1. IT jobs: How much should a CIO, CTO, CISO earn in 2017?

    IT jobs: How much should a CIO, CTO, CISO earn in 2017?

    Chief information officer:  Salary growth 1.8%

    2016 salary expectations: £126,500 - £243,250   2017 salary expectations: £128,250 - £248,000

    Chief technology officer: Salary growth 1.9%

    2016: £83,500 - £158,750  2017: £85,250 - £161,500

    IT director: Salary growth 0.5%

    2016: £93,750 - £126,750  2017 £94,000 - £120,750

    IT manager/head of IT: Salary growth 3.3%

    2016: £80,250 - £115,500   2017: £81,500 £120,750

    Chief Information Security Officer: Salary growth 4%

    2016: £98,250 - £149,500  2017: 104,250 - £153,500

    2. IT jobs: How much should an enterprise architect earn in 2017?

    IT jobs: How much should an enterprise architect earn in 2017?

    Chief architect: Salary growth 1.6%

    2016: £120,000 - £155,000   2017: £121,500 - £158,000

    Enterprise architect: Salary growth 1.4%

    2016: £62,750 - £93,750  2017: £63,000 - £95,750

    Technical architect:  Salary growth 2.4%

    2016: £82,250 - £103,250  2017: ££85,400 - £104,500

    Infrastructure architect: Salary growth 3.5%

    2016: £57,500 - £76,750   2017: £60,500 - £78,500

    Data Architect: Salary growth 1.1%

    2016: £59,750 - £79,250   2017: £61,000 - £79,500

    Solution Architect: Salary growth 0.9%

    2016: £58,250 - £81,500   2017: £58,250 - £82,750

    3. IT jobs: How much should developers earn in 2017?

    IT jobs: How much should developers earn in 2017?

    Senior developer: Salary growth 0.9%

    2016: £53,750 - £88,750   2017: £54,000 - £89,750

    Developer: Salary growth 4.5%

    2016: £26,750 - £45,500   2017: £27,500 - £48,000

    Junior developer: Salary growth 4.2%

    2016: £19,000 - £28,500  2017: £19,500 - £30,000

    Database/business intelligence developer: Salary growth 2.9%

    2016: £41,500 - £69,250  2017: £44,500 - £69,500

    4. IT jobs: How much should IT testing professionals earn in 2017?

    IT jobs: How much should IT testing professionals earn in 2017?

    Test programme manager: Salary growth 2.7%

    2016: £62,250 - £72,750   2017: £65,750 - £76,000

    QA/testing manager: Salary growth 1.4%

    2016: £50,750 - £70,250   2017: £51,000 - £75,000

    QA/testing analyst: Salary growth 2.5%

    2016: £23,500 - £45,250   2017: £24,500 - £46,000

    5. IT jobs: How much should project managers & business analysts earn in 2017?

    IT jobs: How much should project managers & business analysts earn in 2017?

    Programme manager:  Salary growth 2.6%

    2016: £62,250 - £99,750   2017:  £65,750 - £100,500

    Project manager: Salary growth 1.2%

    2016: £39,250 - £68,250    2017: £39,750 - £69,000

    Senior business analyst: Salary growth 1.9%

    2016: £45,250 - £58,750  2017: £46,500 - £59,500

    Business analyst: Salary growth 1.2%

    2016: £34,250 - £49,250  2017: £35,000 - £49,500

    6. IT jobs: How much should data analysts earn in 2017?

    IT jobs: How much should data analysts earn in 2017?

    Data analyst: Salary growth 2.2%

    2016: £32,500 - £46,250   2017: £32,750 - £47,750

    Business intelligence analyst: Salary growth 3.1%

    2016: £42,750 - £54,250   2017: £44,000 - £56,000

    7. IT jobs: How much should infrastructure and engineering professionals earn in 2017?

    IT jobs: How much should infrastructure and engineering professionals earn in 2017?

    Infrastructure manager: Salary growth 1.6%

    2016:  £48,500 - £75,250  2017: £50,250 - £75,500

    Database administrator: Salary growth 0.5%

    2016: £37,500 - £66,750  2017: £36,000 - £67,000

    Network manager: 2.8%

    2016: £38,250 - £58,750   2017: £39,250 - £59,500

    Network engineer: Salary growth 2.8%

    2016: £35,750 - £53,750   2017: £38,000 - £54,000

     

    8. IT jobs: How much should IT support & operations staff earn in 2017?

    IT jobs: How much should IT support & operations staff earn in 2017?

    Support manager: Salary growth 3.0%

    2016: £39,500 - £51,250  2017: £40,000 - £53,500

    Systems/virtualisation engineer: Salary growth 0.8%

    2016 £38,750 - £50,750  2017: £39,000 - £51,250

    Systems administrator: Salary growth 1.7%

    2016: £21,250 - £37,750   2017: £21,500 - £38,500

    Desktop, first/second line support: Salary growth 1.5%

    2016: £21,250 - £28,500 2017: £21,750 - £28,750

    Applications/production support manager: Salary growth 0.7%

    2016 £62,500 - £77,250   2017: £63,000 - £77,750

    Senior applications/production support: Salary growth 2.6%

    2016: £39,500 - £58,500  2017: £42,000 - £59,500

    Applications/production support analyst: Salary growth 2.4%

    2016: £31,250 - £41,500   2017: £32,500 - £42,000

    9. IT jobs: How much should compliance, risk & security professionals earn in 2017?

    IT jobs: How much should compliance, risk & security professionals earn in 2017?

    Information security manager - Salary growth 1.9%

    2016: £63,250 - £97,500   2017: £65,000 - £98,750

    Information security officer -  Salary growth 2.2%

    2016: £45,750 - £ 68,250 2017: £46,250 - £70,250

    Security network engineer - Salary growth 2.9%

    2016: £33,500 - £ 52,500 2017: £34,750 - £53,750

    10. IT jobs: How much should digital, ecommerce and mobile developers earn in 2017?

    IT jobs: How much should digital, ecommerce and mobile developers earn in 2017?

    Mobile Applications Developer:  Salary growth 4.3%

    2016: £32,500 - £ 61,500  2017: £36,000 - £62,000

    Senior Web Developer: Salary growth 3.4%

    2016: £51,750 - £66,750   2017: £55,500 - £67,000

    Web Developer:  Salary growth 2.2%

    2016: £36,500 - £55,000   2017: £36,750 - £56,750

    Web Administrator:  Salary growth 1.5%

    2016: £25,750 - £42,750  - 2017: £26,000 - £43,500

    Web designer: Salary growth 1.3%

    2016: £27,500 - £47,250   2017: £28,250 - £47,500

    • Digg
    • Del.icio.us
    • StumbleUpon
    • Reddit
    • RSS

    Five real-life use cases for Google DeepMind’s machine learning systems

    DeepMind has attracted mixed headlines since Google paid £400 million for the artificial intelligence (AI) startup in January 2014. When an AI system built by DeepMind won a series of Go matches against the world's best player of the Chinese board game, the possibilities of the technology inspired wonder. But that was tempered by criticisms of its controversial access to personal health records, with concerns raised over data privacy.

    From their lab in Kings Cross, London, 140 DeepMind researchers develop software systems capable of independent learning through AI algorithms. DeepMind is working with a small number of organisations to implement its technology so far, but there are already some intriguing real-life examples emerging.

    Read next: Google DeepMind: What is it, how does it work and should you be scared?

     

    1. DeepMind use cases: Royal Free NHS Trust

    DeepMind use cases: Royal Free NHS Trust
    Image: iStock

    In February Google launched DeepMind Health, with the goal of using its machine learning systems to improve healthcare treatment and digitise processes.

    The first project announced is a partnership with the Royal Free NHS Trust in north London, involving the creation of a mobile app that provides clinicians with cutting-edge analytics. Known as Streams, the app was designed to improve the detection of acute kidney injury by instantly reviewing blood test results before sending an alert to the most appropriate clinician via a handheld device.

    The data-sharing agreement gives access to encrypted healthcare data from an estimated 1.6 million patients who use the three hospitals run by the trust annually: the Royal Free, Barnet, and Chase Farm. The project attracted controversy for providing Google with confidential patient information, despite assurances from the tech giant that the information would only be used to inform diagnosis and treatment.

    Read next: Google’s DeepMind promises openness as it begins public consultation over healthcare plans

    2. DeepMind use cases: Moorfields Eye Hospital

    DeepMind use cases: Moorfields Eye Hospital
    Image: iStock

    The second NHS collaboration announced by Google was a partnership with Moorfields Eye Hospital. The project was set up to develop a machine learning–based system that can recognise sight-threatening eye diseases from a digital scan of the eye. While its initial collaboration with the NHS at the Royal Free focused on patient care, this is the first that is entirely dedicated to medical research.

    The programme involves the analysis of more than one million anonymous eye scans to produce an algorithm that detects early signs of emerging eye conditions and increases the speed of diagnosis.

    A Moorfields ophthalmologist called Pearse Keane was credited with first introducing the idea. He contacted the company after seeing their technology help computers learn how to play video games and believed it could be applied to images of the eye.

    3. DeepMind use cases: University College London Hospital

    DeepMind use cases: University College London Hospital
    Image: iStock

    DeepMind's latest research partnership with the NHS aims to tackle head and neck cancers. Before radiotherapy can begin, clinicians currently spend around four hours preparing a detailed map of each patient's body to avoid targeting the delicate surrounding tissue that can be damaged in treatment. The information is then fed into a radiotherapy machine to target the cancer without harming the healthy tissue.

    Researchers at DeepMind believe machine learning can cut this time down to an hour. The team will analyse anonymised scans from UCLH patients to develop a radiotherapy segmentation algorithm that can automate parts of the process. They hope to eventually apply the algorithm to other parts of the body.

    4. DeepMind use cases: WaveNet

    DeepMind use cases: WaveNet
    Image: iStock

    While healthcare technology dominates the current DeepMind developments, its machine learning systems have also been extended to audio. Talking machines have a long history in science fiction and have already entered mainstream usage through commercial products such as Apple's personal voice assistant Siri. But the gap between __computer and human speech remains substantial.

    DeepMind has developed a text-to-speech system that can close that gap by more than 50 per cent. Known as WaveNet, it uses a neural network to replicate the soundwaves produced by human speakers rather than the copying the language they use. The computing power required to power Wavenet limits its practical applicability for now, but Google has already presented samples of its automatically generated piano pieces.

    5. DeepMind use cases: Google

    DeepMind use cases: Google
    Image: iStock

    Google already uses machine learning algorithms in a range of its own products, including Maps, Gmail, YouTube and Android. It believes that DeepMind’s technology could also be extended to search, robots and the internet of things in future. A DeepMind agent has already matched human performance at 49 Atari games including Pac-Man and Space Invaders, and developed the aforementioned AlphaGo, the first __computer programme to ever beat a professional player at Go.

    Google has even used DeepMind to cut its electricity bills at its huge data centres. DeepMind algorithms were used to predict the air conditioning required to cool the vast number of servers powering its services, which can vary depending on the user demand. The results were efficiency savings of 40 percent in the cooling systems along, and a reduction to the overall energy used in the data centres of 15 percent.

    • Digg
    • Del.icio.us
    • StumbleUpon
    • Reddit
    • RSS

    Telefonica creates 'unicorn team' of data experts to standardise business intelligence practices internationally

    Telefonica has created a specialist team of data experts to help standardise business intelligence (BI) practices across its South American operations.

    The telco giant operates on a federated model, with each international division responsible for its own organisational structure and technology choices. Business intelligence (BI) has always been the domain of marketing teams in each geography, making it hard to create a consistent analytics strategy across the business.

    To solve this, the company created what it calls a global “unicorn team”, combining data scientists with complimentary roles such as data journalists (who visualise and present data), data wranglers, data engineers and line-of-business staff to deliver their own data solutions. The term comes from the idea that data scientists are as rare as unicorns.

    This central data team allows the company to create a standard approach to BI across 19 countries in South America countries, John Belchamber, global leader of advanced analytics at Telefonica told Computerworld UK. 

    This blended team assist in the deployment of BI software tools that were previously prescribed by the individual IT departments, an approach which slowed new projects in some cases.

    This new approach allows each country data analytics efforts "to be more quickly monetised," by surfacing a more granular view of customer habits in order to deliver more targeted marketing, and to be more reactive in order to reduce customer churn.

    Read next: Telefonica CIO: Tackling legacy IT 'burden' key to battle with Google and WhatsApp

    Belchamber's team then provides on-line training and a community for the global teams to share best practices once they are up and running.

    Telefonica allows each business unit to make their own technology choices, as long as the tools are being employed to a specific end. "We are open source, we take the business goal first and let these teams choose the tools they need to get there," he explains.

    Read next: Ten of the best self-serve analytics and business intelligence tools for enterprises: What are the best alternative BI products?

    The overarching aim of all of this is the same as most organisations: to understand the customer better. The key to this is better customer profiling. The BI team is taking a few different approaches to this, from analysing contact centre call records, to parsing web browsing data into a customer profile

    In terms of web-profiling Belchamber's team looks to give each business unit a 'dictionary', which is essentially a piece of bespoke-BI software that parses network data -- so web pages visited, when and on what device -- and builds a media profile of that customer to be used by the local team to deliver targeted marketing.

    As a telco Telefonica looks at mobile browsing history or fixed line browsing. Naturally all of this is done within the legal framework of the geography that the technique is deployed, looking at customers that have opted-in, or using anonymised data.

    Read next: Pret a Manger separates BI systems from core IT to power real-time reporting

    Telefonica has also invested in speech and natural language processing (NLP) to discern sentiment from call centre chat logs. Unfortunately because of data restrictions the unicorn teams couldn't analyse the actual speech of the logs, so they anonymised the MP3 data and looked at the background acoustics. Belchamber says this still allowed them to analyse the "heartbeat of the call".

    "We found within Latin American culture, who might be more expressive, there are peaks and troughs from which we can predict a good or bad call". Belchamber says this is the "first stage to understand customer journeys and introduce history into those transactions".

    The problem is that most existing NLP APIs, such as Microsoft's, are built for English speakers, so the organisation has had to start from scratch. "We are now building true speech analytics and NLP pieces," Belchamber says.?

    • Digg
    • Del.icio.us
    • StumbleUpon
    • Reddit
    • RSS