Smarter software is coming … just slowly
Smart software is in demand. Large quantities of text should make sense to an indexing system. The specific needs of a knowledge worker should be able to personalize each information task. Instead of a knowledge worker scanning images or videos for important facts, a gizmo in a data center should do that work. The weasel word is should. Today’s systems are getting incrementally more intelligent at what is a snail’s pace compared to advances in mathematics.
Mathematicians are making remarkable strides. The Fields Medal winners for 2014 provide a window into the fast-moving world of theoretical and applied numerical recipes. (A list of the winners can be viewed at mathunion.org/general/prizes/2014.) The Fields Medal is the equivalent of an Olympic gold medal and a Nobel Prize. The four medal winners are the tip of the math iceberg. Hundreds of other gifted mathematicians are pushing the boundaries in many areas of mathematical interest.
How will the mathematicians’ work improve knowledge management, information retrieval and analytics that deliver high-confidence results? First, think in terms of years. The Field winners’ ideas will filter to their students and to colleagues. Students who wrestle with those methods will discover or “see” applications to help solve specific real-world problems. Once those insights have sparked an individual, it may be many years before an application of the methods finds its way into a Google search system or provides the secret sauce for an information processing company. Next, consider the time scale. The go-to numerical recipes for search and query processing make use of methods that are in many cases centuries old.
The possible applications of this year’s Field Medal winners are, of course, almost impossible to predict. Moving mathematical innovations into a shipping product is always a challenge. The math procedures face another hurdle as well. Certain combinations of procedures impose a computational burden on the systems used to apply the procedures. Academics refer to operations that are non-computable. Just because today’s systems cannot run the procedures does not mean the numerical recipes are wrong. The problem is with the step-by-step approach of computers. Google and many other companies are pumping money and talent into alternatives to the von Neumann approach we take for granted.
The hardware factor
Beyond mathematics, there is the need for hardware. IBM announced in August that it has invented a “new computer chip that thinks like a human,” according to the Motley Fool. (See fool.com/investing/general/2014/08/13/ibms-new-computer-chip-thinks-like-a-human.aspx.) The IBM Research explanation of its neurosynaptic chips states: “IBM built a new chip with a brain-inspired computer architecture powered by an unprecedented 1 million neurons and 256 million synapses. It is the largest chip IBM has ever built at 5.4 billion transistors, and has an on-chip network of 4,096 neurosynaptic cores.” (See research.ibm.com/cognitive-computing/neurosynaptic-chips.shtml#fbid=pNpS67Gy6av.) The goal of the new chips will be to “create a holistic computing intelligence.” The new chips “address the senses and pattern recognition.”
In that IBM research, Dharmendra Modha, an IBM Fellow, said: “The architecture can solve a wide class of problems from vision, audition and multisensory fusion, and has the potential to revolutionize the computer industry by integrating brain-like capability into devices where computation is constrained by power and speed.”
In contrast to an old-style Intel chip, IBM’s neurosynaptic chip has one million programmable neurons, 256 programmable synapses and 4,096 neurosynaptic cores. IBM revealed: “The neurosynaptic chip veers from the traditional von Neumann architecture, which inherently creates a bottleneck limiting performance of the system.”
As part of its commitment to smart software, IBM announced in September 2013 that Watson Analytics will get “analytics in the hands of every business user, which is a real challenge.” (See bits.blogs.nytimes.com/2014/09/16/ibm-offers-a-data-tool-for-the-mainstream-with-watsons-help.) As advanced as Watson is, the system makes use of well-known computational methods. Presumably IBM’s innovations will eventually find their way into products and applications.
A creative problem
Google is active in advanced computing as well. The company has its own Quantum Artificial Intelligence Lab. Hartmut Neven, the director of engineering for the Google AI activities, said: “Machine learning is highly difficult. It’s what mathematicians call an ‘NP-hard’ problem. That’s because building a good model is really a creative act. As an analogy, consider what it takes to architect a house. You’re balancing lots of constraints—budget, usage requirements, space limitations, etc.—but still trying to create the most beautiful house you can. A creative architect will find a great solution. Mathematically speaking, the architect is solving an optimization problem and creativity can be thought of as the ability to come up with a good solution given an objective and constraints.”
Neven continued, “Classical computers aren’t well-suited to these types of creative problems. Solving such problems can be imagined as trying to find the lowest point on a surface covered in hills and valleys. Classical computing might use what’s called ‘gradient descent’: Start at a random spot on the surface, look around for a lower spot to walk down to, and repeat until you can’t walk downhill anymore. But all too often that gets you stuck in a ‘local minimum’—a valley that isn’t the very lowest point on the surface. That’s where quantum computing comes in. It lets you cheat a little, giving you some chance to “tunnel” through a ridge to see if there’s a lower valley hidden beyond it. This gives you a much better shot at finding the true lowest point—the optimal solution. (See googleresearch.blogspot.com/2013/05/launching-quantum-artificial.html.)