What is DARPA?
DARPA is an acronym that stands for "Defense Advanced Research Projects Agency", or agency for advanced research projects for Defense.
It is an agency created in the United States in the 1958, behind the thrust of the cold war and the overcoming of space research carried out by Russia with the launch of the Sputnik 1. The agency was initially called ARPA and only some time later it joined the national defense and became DARPA.
Among the programs developed, here are some that are now known to all:
- HAVE BLUE project: in the early seventies it led to the creation of new stealth technology aircraft, the F-117;
- the miniaturization of the Global Positioning Sistems (GPS), carried out in the eighties, allowed the development of the GPS that we know today, integrated in every kind of portable and low consumption device;
- ARPANET (Advanced Research Projects Agency NETwork) was developed as a computer network for DARPA support and is, in practice, the network that gave birth to the Internet.
Every year DARPA finances research and study programs in the most diverse fields of knowledge.
One of the projects that has led to recent successes is known as SyNAPSE Program Develops Advanced Brain-Inspired Chip. Through the SyNAPSE (Systems of Neuromorphic Adaptive Plastic Scalable Electronics) program, IBM researchers have created the most complex Chip inspired by the operating principles of the brain structure.
The chip, called TrueNorth, contains more than 5 billions of transistors and allows you to program 256 millions of programmable logic points, equivalent to synapses. Of course, still very far from the number of synapses of the human brain, but still a respectable result. TrueNorth is a new type of processor, built according to a new architecture, which was developed starting from the most advanced knowledge on the structure and functioning of the brain.
Among the "fathers" of the new processor we find Dharmendra Modha, a IBM Fellow, one of the most important IBM scientists, head of the project Cognitive Computing.
The discipline of Cognitive Computing is that which deals with studying and solving problems of high complexity, characterized by ambiguity and uncertainty, problems that are normally "solved" by human beings, is the case of the recognition of a face, for example.
For the IBM a system of Cognitive computing it carries out the following tasks: "to learn and interact naturally with people or to take their own hands. They help human experts make better decisions by penetrating the complexity of Big Data".
But why all this interest in building a machine that has capabilities that men already have?
For the same reason that the computers we already know were born. The man was able to count even before the arrival of the computers, but if the operations are repetitive and long it is difficult to find someone who takes care of them and who, perhaps, does not make mistakes.
Here is a computer, optimized to make calculations, takes the place of one or more men, who instead of making calculations worry about writing instructions for the computer and verifying that it works well.
Similarly, if we have to check millions of faces on a monitor and compare them with cards to verify the identity of people, we can put 100.000 people behind the monitors to verify who is passing under us, or we can create software that, by performing a very large number of simple operations, they are able to "recognize" people already on file. This has already been done. there are many recognition systems in operation that are based on computers built according to the von Neumann architecture.
It is easy to realize, if you have ever entered a data center, the enormity of the resources needed to do certain operations and immediately catch the eye that classic computers, while very fast in doing simple repetitive tasks, are very inefficient in the find solutions to complex problems, such as the one shown above. Inefficiency reflects on the resources needed to perform the task, resources that we can measure in various ways, one of which is energy consumption.
A computer, from this point of view, is much less efficient than a brain.
This is where the idea of building new computers with a brain-like structure comes from.
The combination of classic computers and computer type cognitive, could be able to create new super computers able to quickly perform repetitive calculations and also to quickly recognize in the person next to us, the face of a dear friend or an enemy. The interest of American military circles in this new technology is immediately understandable.
For now we try to better understand how it is done, what new fields of research and study, even in a military environment, which opens the new IBM processor. In fact, the new technology needs new programming languages, new memory logics ... in short, a new world is opening up to engineers and computer scientists, a new world that needs capable and curious explorers.
We will see what the near future awaits us.
