NVIDIA CEO Jen-Hsun Huang’s GTC 2013 keynote was a typical whirlwind tour (with real wind, but that’s a different article) through all the various GPU-related worlds that NVIDIA is touching these days. These addresses are usually chock-full of demonstrations showing where we are in terms of state-of-the-art graphics, scientific and technical computing, entertainment, and now: finding dresses.

In this demonstration, Jen Hsun leafed through the latest edition of In Style magazine. While the models are svelte (or starved), the magazine definitely isn’t, weighing in with 594 pages of ads. A dress from one of those ads was chosen, its picture was taken, and it was sent off for image matching. What came back was a set of likely matches that the image-matching tool found via eBay. (This can be seen in the semi-blurry picture taken from my third-row perch.)

Hmm… now that I think about it, this technology probably isn’t confined only to dresses. With some minor technical tweaks (like checking different boxes), I imagine it would be quite possible to match many other items. I’m thinking handbags, blouses, shoes, skorts, and even jorts for those needing to feed their denim demons.

They also demonstrated that it’s possible to capture a particular pattern and then search for clothing that has the same, or a similar, look. To my untrained eye, it looked to do a pretty good job. It didn’t find exact matches, but the selection shown came pretty close to the mark.

The impressive thing about this tool is its accuracy and speed. On each demo it not only returned the correct type of garment, but the results were surprisingly close to the original image in terms of look and general configuration. And it took only a few seconds – not much longer than the loading time for a web page.

There are already a fair number of images on the Internet, and users of Facebook add something like 300 million more per day. On the video side, there’s something like 72 hours of video added to YouTube per minute. Over time, this is going to add up. There will be an acute need for more sophisticated image searching/matching technology.

So – aside from everyone who likes to shop for clothes, who will use this technology? The companies who want to make it quicker and easier for potential customers to comb through their vast inventories of goods. With our increasing reliance on communicating via images, the ability to search, sort, and match is going to become more important over time.

The post GPUs Power Big Data for Frock Finding appeared first on Inside-BigData.

Procter & Gamble is going to quadruple the company’s staff of business analytics experts in the near future. This is despite the fact that the company is reducing spending in other categories, including significant non-manufacturing layoffs and a hefty 30% cut (equaling $1 billion) in annual IT spending.

So why would P&G look to build their analytics ranks so large, so quickly? It’s because they see a new business model emerging, and they’re embracing it.

As P&G CIO Filippo Passerini explained in a recent Information Week article, traditional IT and analytics focused on delivering the right data and reports as soon as possible (usually days or weeks) to key analysts and decision makers. This, of course, is the tried and true model that vast majority of companies are still using today.

What Passerini sees coming is a new model that relies on ad hoc teams of subject matter experts who would assemble, virtually or physically, to address problems the moment they arise. These teams would pull together to handle a specific situation and then dissolve when a solution is reached.

This model is dependent upon a fast and deep flow of granular, real-time data. P&G estimates that right now, they have about 60% of the real-time data they need, and they’re busily working to get access to the rest of it. Much of this data is the typical stuff you’d expect like point-of-sale numbers, manufacturing output, etc. but it’s delivered much faster than is normal for most firms.

In the article, Passerini discusses what Procter & Gamble is trying to figure out in broad terms. They’re ahead of many companies in getting to what I call “One version of the truth,” meaning that everyone agrees on what metrics to use and how to apply them.

P&G is mainly focusing on the next step: given a ‘truth,’ for example, a drop in sales for a particular product, their job is to figure out why. It could be because the overall market has shrunk, or a competitor is taking share, or inventory issues, or a wide variety of other reasons. P&G is working on automating the ‘Why’ piece by alerting product managers to things like competitor moves or production problems.

P&G is hiring analytics pros mainly for the final step – deciding what to do to fix a particular situation. They’re looking for people who will live where IT and business meet. These are folks who understand P&G’s products and markets, and the dynamics of both consumers and competitors.

But they are equally comfortable building models and simulations in order to better understand the existing conditions and the most important factors. With this information, they can recommend the best course of action for solving any P&G problem.

The key to this is speed, speed, and, yes, more speed. In the P&G model, answers and solutions need to be immediate. They won’t take weeks to gather information and formulate ‘what ifs’. In P&G’s world of Big Data, the key players have the data they need to accurately assess the problem and apply the right prescription to fix it or at least improve the situation.

Making the right move at the right time is as important to P&G’s efforts at selling soap as it is to any Wall Street firm trading a multi-billion dollar portfolio – and now firms like P&G are staffing up with the same analytic experts that Wall Street has been hiring for years.

The post Big Data for Big Careers – P&G Quads Down on Analytics Expertise appeared first on Inside-BigData.

When HP announced its purchase of UK analytics firm Automony for $10.4 billion back in September 2011, it shocked HP shareholders and caused many industry insiders to comment that HP was paying way too much for the company.

At the time, Autonomy was a 15-year-old analytics firm that specialized in helping enterprises manage all types of data, with an emphasis on pan-enterprise search and dealing with the growing problem/opportunity inherent in the explosion of unstructured data.

Pre-purchase, Autonomy boasted high recessionary revenue growth of 18% in 2010, solid 43% operating margins, an average deal size of $800,000, and a customer list that included some of the largest high-profile enterprises in the world. But the company still hadn’t topped a billion in revenue and certainly didn’t have the profile of a SAS or SPSS.

HP paid a 79% premium based on Autonomy’s pre-acquisition share price. So what has HP gotten for their money? Did they get hornswoggled, or did they get a good deal on a hidden gem?

A trio of recent articles take a look at what Autonomy brings to the HP table. The first is a Los Angeles Times piece about how Autonomy’s technology is being used to bring ‘augmented reality’ to the masses – allowing devices to understand what they’re seeing and act upon that information. One of the examples from the article discusses how a consumer might point their phone at a jar of pasta sauce and automatically get wine recommendations. (I’ve always wondered which box of wine would best complement my Chef Boyardee Beefaroni.)

CRN interviews former Autonomy founder-CEO (and current HP VP) Mike Lynch, asking how the company fits into HP and, specifically, what value the company might bring to HP channel partners. A short piece from Business Insider takes a bit of a skeptical look at the company and how it’s positioned for future growth.

The jury is still out on whether Autonomy was a good deal for HP. They’re now rolling out Autonomy-infused hardware appliances along with software/service packages, showing cooperation and cross-fertilization between the companies. Autonomy already had than 60,000 customers and, as was outlined in the Business Insider piece, has closed some huge deals with equally huge customers, so HP is getting the benefit of that business.

The addition of Autonomy will almost certainly help HP get a larger wallet share with their joint customers as sales of software secret sauce drags along HP hardware and services. This alone might be enough to justify the purchase price.

To me, HP had to make a move to stake out a claim in the big data gold rush. There isn’t time to organically build their chops in big data; the market is moving much too fast. So HP bought their way in, just like both IBM and Oracle have done.

HP’s purchase of Vertica got them part of the way there, but Autonomy’s advanced and even unique capabilities give HP the credibility and tech heft to compete for the largest and most ambitious big data deals. With Autonomy (and Vertica too), HP can put together custom large-scale solutions for their biggest customers and, at the same time, develop dedicated data appliances that will be point solutions for everyone else.

It could be a compelling value proposition, and one that puts HP squarely in the same league with IBM and Oracle – assuming that they can properly craft and execute on the strategy.

The post HP Hornswoggled? Will Autonomy Purchase Pay Off? appeared first on Inside-BigData.

One of the areas where “Big Data” will have the most impact is in health care. Applying analytics to medical research and treatment will extend human lives and improve the quality of life for, well, pretty much everyone. Recently, health insurer WellPoint announced that it will be the first commercial customer for IBM’s Jeopardy-winning Watson system.

A Watson system optimized for health care would have the ability to understand patient symptoms and conditions and then scan every medical textbook, journal, study, and clinical trial to suggest possible courses of action.

The system won’t be making decisions; it will be more like the Data character on Star Trek: The Next Generation. Imagine a system that can quickly and accurately produce a list of potential conditions, suggest tests that will winnow the list down, and then offer up advice on the treatment that offers the highest probability of success while taking into account the unique characteristics of each patient.

The payoff will be substantial. It should radically decrease the number of times doctors misdiagnose a condition, and it will ensure that they are using the best treatments available. It will also eliminate much of the wasteful testing and “defensive medicine” that pushes up health care costs while adding little value for patients.

It always surprises me how individual doctors will treat the same conditions in quite different ways. They have the power to use leeches or lasers. How any doctor approaches a condition is based on his/her experience and, to a large extent, the latest research they’ve heard about.

Most of the time this is OK, but in some cases it can result in the wrong treatment and bad outcomes. With analytic-based assistance, doctors will see which tests are necessary and which won’t add any value saving time, money, and patient distress.

Analytics will also help researchers figure out which treatments produce the best outcomes. The head of a very large medical organization once told me that if he could give researchers a simple database that contained millions of records covering diagnosis, treatment, outcome, and patient history (no names), they would be able to radically improve treatments for every condition.

Unfortunately, we’re a long way from analytics aiding research as I’ve outlined above. It’s not a technical problem; it’s a set of cultural, political and economic hurdles that have frozen the existing system in place.

Medical vendors (particularly ISVs) are loath to change their proprietary coding schemes. Patients don’t want to give access to their health records to anyone for fear it could be used against them. Insurers (and the government too) want access to the records for their own reasons which is a scary prospect for many of us.

But progress is being made on all of these fronts, and analytics-led medicine is becoming a reality. In the next five years we should see an explosion of systems, software, and even new treatments that have come about because of this new focus on analyzing data.

Dan Olds is CEO of Gabriel Consulting Group and Chief Editor of inside-BigData.

The post Big Data & Doctors: Analytics-led Medicine is Coming appeared first on Inside-BigData.

The big business news this week is how a junior trader making unauthorized transactions cost banking giant UBS at least $2.3 billion in losses. The loss tally is still mounting as the bank works to figure out exactly how big a hit they’re going to take.

Given that UBS’s market capitalization is around $42 billion, and net income was around $7.5 billion in 2010, a $2.3 billion loss is more than just a haircut; it’s closer to a scalping.

So how could something like this happen? Easier than one would like to think. In this situation, the trader was taking speculative positions on index futures – simply put, he was betting on where the markets (S&P 500, DAX, EURO STOXX indexes) would be at some future date.

While it was OK for him to make these bets, he was supposed to place offsetting bets that would limit the risk exposure of the bank. So if he took a position betting that the S&P would be higher at a future date, he was supposed to hedge that position with another transaction to ensure that the bank didn’t take it in the shorts if he was wrong.

Did our guy do this? Noooo… he faked the offsetting transactions; his trades only looked as if they were nicely balanced. In reality, he exposed the bank to huge amounts of risk – bad dreams that seem to have come true in recent days.

There was a big hole in the UBS trader management and risk control systems; that’s pretty obvious. The best corporations have risk management mechanisms that allow management to see organization-wide exposure at any given moment and how changing conditions could make the picture better (or worse).

Designing and implementing this kind of system isn’t trivial, particularly for a multi-national corporation. Accurately analyzing the data and providing up-to-the-minute real results – and predictions of future results – is even more difficult.

But all of this highly elegant (or brute force) programming and analytics are wasted if the overall solution has huge holes in it. If the traders can enter false trades, then the system can’t be trusted – as UBS has discovered.

And UBS is not alone. Bernie Madoff was able to issue false statements and pass regulatory muster for decades. The models used by ratings agencies to rate mortgage debt didn’t take into account systemic mortgage default risk… the list goes on and on.

The bottom line is that the most sophisticated systems can be rendered worthless by low-level, basic oversights – like allowing false trades to masquerade as real transactions. The lesson is to poke holes in your analytic systems and question even the most basic assumptions. To me, it’s much better to be known as ‘the guy who asks the dumb questions’ than ‘the guy who is now a cautionary tale’.

Dan Olds is CEO of Gabriel Consulting Group and Chief Editor of inside-BigData.

The post Rooting out Rogues – The Role of Analytics appeared first on Inside-BigData.

Dan Olds

So what is ‘Big Data,’ and why should you care about it? To take the first question, well, first: the specific term ‘Big Data’ refers to the radically increasing size of data sets that need to be gathered, moved, and analyzed by research, government, and business organizations. In a more general sense, the Big Data trend is about organizations of all types and sizes finding that their ability to gather, analyze, and act upon available data increasingly determines whether or not they achieve their particular mission.

For traditional research organizations (national labs, academic institutions) Big Data is what they’re working with on a daily basis – although even their data sets are growing at a much faster pace than many envisioned. Likewise, many private organizations in energy production, life sciences, financial services, and even retail have also been analyzing massive data sets and using the results to guide their actions and strategies for many years. Now we’re seeing this trend become broader and deeper as organizations that are already doing ‘big analytics’ extend their capabilities, and newcomers start to dip their toes into the pool.

What’s driving this? Why is it happening now? To me, it comes down to economics and societal changes. At the most basic level, globalization and the instantaneous communications enabled by the wired economy have put almost all of the economic power into the hands of buyers.

Think about how you purchase goods and services today. Do you walk into a store or two and make your buying decision based on what a salesperson or store display tells you? Or do you research the alternatives online – examining features/functions, looking at product and vendor reviews, and finding the best value for your money?

I’m firmly in the latter camp, rigorously shopping online for every significant purchase. Business buyers are behaving the same way when looking for suppliers or even employees. They search for the best combination of quality, terms, and price from suppliers who are located anywhere in the world. Globalization, with the lowering of trade barriers worldwide, makes it much easier to manage a global supply chain or set of end-user customers.

In this kind of economic environment, it’s very hard to achieve and maintain a competitive advantage over your rivals. Successful new innovations on the product or service front are quickly adopted by competitors. Your market can suddenly be disrupted by new entrants or entirely different offerings that threaten to make your product or service obsolete. The end result is a constant churn that keeps margins low and makes it hard for a company to break away from the pack for any significant period of time.

The use of analytics and predictive analytics is fast becoming a valuable tool that enterprises are using to cut costs and maximize opportunities. Virtual prototyping allows Boeing to design and test an airplane wing without having to physically build it first. Wall Street firms use analytics to build models and trading algorithms to maximize their market returns. Cities can use analytics to deploy the level of police presence they need to reduce crime in trouble areas. One very large retailer uses analytics to track cold and flu season as it moves across the world so that they don’t need to buy more tissue and cough drops – they just make sure to have the right quantities in the right locations.

Many companies may think that they’re already doing Big Data (or, more accurately, enterprise analytics) because they’re using Business Intelligence (BI). From what I can tell, many of them aren’t really in the same zip code yet. Their BI systems, while impressive in many cases, just don’t make the cut in terms of providing true insight and actionable intelligence. The biggest downfall is that these BI operations rely heavily (almost exclusively, in many cases) on data generated by the organization itself. Thus predictions of future conditions are built primarily on past experience – past experience that’s limited to a single organization.

This is a lot like driving down the freeway with your eyes locked on a very large and shiny rearview mirror. Your steering decisions are based on landscape that just rushed by your window. This worked fine for the most part in “normal” times (what are those?) But today curves, washed-out bridges, and mountains materialize much more quickly. In order to predict these changes, or to at least figure out how you should react to them, you need to have a much wider and larger set of data inputs to analyze.

The processing required to discover, quantify, and predict business conditions isn’t all that different than what HPC researchers are doing daily. The data and the questions being posed are different, of course, but the statistical techniques and the math are the same. The IT infrastructure supporting both efforts will be much the same as well. However, many corporate data centers will need to consider how (or if) their existing infrastructure can address more computationally intense workloads and deal with much larger datasets.

Inside BigData will track this trend as it develops, providing news on a wide variety of topics including both technical and business innovations. We’ll evolve as the trend evolves, but the focus will always be on providing our readers with interesting and timely content that will help them get the most out of their data. And feel free to let us know how well we’re hitting (or missing) the target, we appreciate any and all feedback.

The post Big Data Is a Big Deal — And it’s only going to get bigger… appeared first on Inside-BigData.