Big Data is getting Bigger and Bigger

It was the science-fiction author William Gibson who coined the term “CYBERSPACE”, but he did decades before Internet became essential part of our lives: it was 1982. At that time, foreseeing with incredible accuracy what it would have happened many years later, he wrote novels where the real world was under the control of a “virtual world”, made by millions of lines of codes called “programs” (later identified with the more generic term “software”).

He was an incredible, maybe unintentional, prescient of how the virtual world would change our real world and how, now and more and more in the future, is constantly changing the way we live and behave as human beings.

The MATRIX trilogy movies of the Wachowski Brothers, gave all of us a clear evidence of what the “software”, supported by huge amount of computation power with access to immense amount of data, could do: make all of us actors of a movie where our characters, screenplay and direction are decided by a new entity which has nothing to do with any religions we do know or follow.

But this is just a “science-fiction” movie, right? “So why do you mind”, many of you are just thinking?

I have some right reasons! But we have to go back in time.

In the mid 90’s, IBM Deep Blue, a programmers’ program, unbroke the barrier between humans and machines, that, until then, nobody (or maybe very few) thought it would have possible.  IBM Deep Blue beat the world chess champion Gary Kasparov, breaking the barrier of common believe that only “computational power” was not sufficient for beating the “unpredictable human creativity”.

But this was a programmers’ program, made by “human beings” who put a lot of efforts and “creativity” to write millions of millions of lines of code for “teaching” Deep Blue how to analyze and evaluate more than 200 million chess combinations per second. The poor Gar Kasparov, the most rewarded chess champion in the human history, lost miserably.

It was just the begin. It was just the 1996.

But the recipe was discovered: computational power, made of many interconnected machines which had access to a lot of data.

And now, where do we stand? We stand that all main drivers or "ingredients" are converging, making possible the first attempts of simulating "human brain".

These three main drivers (or if you like "ingredients" of a recipe) are:

1) Computational Power:

Moore’s Law, which lasted for many decades in establishing itself as “almost scientific” a “simple rule of thumb” where each 18 months the computational power of microprocessors doubled, is outdated. This pace is now even shorter and faster. In ONE of our smartphone there is more computational power than it was on the computers which were controlling one Space Shuttle Mission in the ‘90’s, at 1/10000th of their cost. And this increase in computational power and a huge reduction of the associated cost is ramping up even steeper in the future

2) Interconnection among computers: from LAN’s to ANN’s – Artificial Neural Networks

With the development of the LAN (Local Area Networks) and WAN (Wide Area Networks), thanks to the implementation of the TCP-IP protocol, computers started to interact among themselves, allowing not only to share data but even more important to share and/or aggregate computational power. This was the really very first “sharing economy” example. Since this first example, almost every business models has been affected and many new ones have surged out. But now we are going even further: engineers, programmers, architect, physicians, mathematic and so for have been struggling in designing computer networks which could emulate, if not even overcome, human brain. Now we have “neural networks”: a neural network is made up of many simple, highly interconnected processing elements that work “in concert”. But, conversely with most servers, pc’s or mainframe, each “single processing elements”, like a brain cell of our brain, takes in data, processes it based on its internal state and produces an output. It is a general-purpose machine, not a program intentionally constructed to carry out a particular task. It requires “to be trained”: without it, this single entity is just an empty slate. But as soon as it is trained, it starts to carry out impressive calculation with an efficiency and effectiveness which improve every time, just like our brain which helps up to do better any task as soon as we do repeat it! But in order to succeed, this intelligent network needs the most important thing: DATA. Exactly, DATA.

3) DATA, BIG DATA, BIGGER AND BIGGER DATA

MIT Engineering Department has developed neural network, which analyzing social activities of students (during a three months’ period before the end of a term) can predict with an high accuracy rate, which classes these students would have chosen the following term.  Another example is that one of a student at Imperial College London who taught a neural network to play chess just by analyzing five million positions from a database of computer chess game. After training, the machine was able to play at very high level and even before of the most sophisticated chess software programs which were written specifically to play chess. But the amazing thig is that this student has written no line of code on how play chess, even the basic rules or strategies! But how is this possible? One answer: DATA! To achieve such impressive feats, neural networks need huge training datasets. And more and more data, even the most negligible, will be necessary in order to make neural networks succeed.  Of course, need vast amounts of computing power well interconnected with high-speed networks.

 

 

 

® FabioPrimerano.com // Tutti Diritti Riservati - All Rights Reserved

He was an incredible, maybe unintentional, prescient of how the virtual world would change our real world and how, now and more and more in the future, is constantly changing the way we live and behave as human beings.

But the recipe was discovered: computational power, made of many interconnected machines which had access to a lot of data.

With the development of the LAN (Local Area Networks) and WAN (Wide Area Networks), thanks to the implementation of the TCP-IP protocol, computers started to interact among themselves, allowing not only to share data but even more important to share and/or aggregate computational power. This was the really very first “sharing economy” example. Since this first example, almost every business models has been affected and many new ones have surged out. But now we are going even further: engineers, programmers, architect, physicians, mathematic and so for have been struggling in designing computer networks which could emulate, if not even overcome, human brain. Now we have “neural networks”: a neural network is made up of many simple, highly interconnected processing elements that work “in concert”. But, conversely with most servers, pc’s or mainframe, each “single processing elements”, like a brain cell of our brain, takes in data, processes it based on its internal state and produces an output. It is a general-purpose machine, not a program intentionally constructed to carry out a particular task. It requires “to be trained”: without it, this single entity is just an empty slate. But as soon as it is trained, it starts to carry out impressive calculation with an efficiency and effectiveness which improve every time, just like our brain which helps up to do better any task as soon as we do repeat it! But in order to succeed, this intelligent network needs the most important thing: DATA. Exactly, DATA.