WHEN ARTIFICIAL INTELLIGENCE RESEARCH AND APPLICATIONS WERE MENTIONED IN THE PAST, ONLY ONE OF THE FLYING LABORATORY STUDIES COME TO MIND.
TODAY, APPLICATIONS THAT MAKE EVERYTHING SMART ARE CALLED ARTIFICIAL INTELLIGENCE. NOW THESE APPLICATIONS THAT ADD THE CLASS OF "SMART" TO ALL CONCEPTS THAT HAVE ARISED IN THE PAST PERIOD, AS A RESULT OF THESE APPLICATIONS TO OUR SOCIAL, SOCIAL AND BUSINESS LIFE INTELLIGENT OBJECTS, SMART SYSTEMS, SMART STATE, SMART STATE, SMART CITY.
Perhaps the most important component, even the backbone, of many countries' digital strategies are artificial intelligence applications. In recent years, artificial intelligence applications are now considered and considered as base technology and strategies are created accordingly. On the other hand, artificial intelligence applications are one of the most controversial technology issues today. The discussions are predominantly based on the ethical boundaries of these practices and the currently lacking legal basis. In general, artificial intelligence aims to understand this by taking the human brain as an exemplary model and from there to find "new inventions" and new applications. It is possible to limit this general definition to some concepts. Even the algorithm related to this area to make it more insightful for the machine industry,It is necessary to take a closer look at three key concepts, machine learning and deep learning.
ALGORITHM IS NOT "INTELLIGENT"
Artificial intelligence is a special application method of the algorithm and sequential problem solving method with the help of computers. The algorithm itself is not "smart". The working conditions of the Deep Blue program, which defeated the world chess champion Kasparov in 1996, were a system of certain and unchangeable conditions. So although it was a perfect system, it wasn't smart. When we talk about smart algorithms, we are talking about self-learning systems and the first application of this is machine learning.
WHAT IS MACHINE LEARNING?
In this system, the data is formed by transferring the metadata created with certain information to the computer. To explain from a classic example, 100 dog pictures are entered into the computer and the information (metadata) that these pictures show dogs is added. The computer program tries to create an example and order from the pictures shown to it over the statistical probability according to the variable function state. He makes a determination as to whether the picture shown is a dog. Currently, systems such as automatic translation and autonomous driving using text and voice are good examples of machine learning. Although such systems are also called artificial intelligence applications, computers in these examples actually evaluate the data they have and reach a conclusion or make a prediction based on high probability detection.The application that is more suitable for the definition of artificial intelligence is deep learning.
THE HUMAN BRAIN IS AT THE BASIS OF DEEP LEARNING One of
the most complex and complicated systems of the machine learning system is called deep learning. The main theme of the system is a kind of artificial neuronal networks system created by taking the human brain as an example. A very comprehensive and large data entry called big data is entered into this system. This big data is not a designed data (there are more designed data in machine learning) and it develops the process by transferring the information needed in different planes to a different plane.
AT THE HEART OF DEEP LEARNING IS THE HUMAN BRAIN
One of the most complex and complex systems of the machine learning system is called deep learning. The main theme of the system is a kind of artificial neuronal networks system created by sampling the human brain. A fairly extensive and large data entry is made into this system, called Big Data. This big data is not designed data (more designed data is available in machine learning), and it improves the process by transferring the information it needs in different planes to a different plane. At this point, we are talking about a multi-purpose and multi-faceted detail analysis capability. Thanks to this ability and as a result of spontaneous processes, surprising results appear. Although this system is quite new, at its initial stage, it can be called an underdeveloped system, despite the ability, flexibility and versatility of human intelligence. Currently, 100 thousand neuronal networks can be built in the most advanced deep learning system, and it is estimated that this number will be more than 10 million in recent history. Applications that will consist of 10 million neuronal network systems will probably surprise us humans even more, but given that the estimated neuronal networks in the human brain are about 85 billion, it can be said that artificial intelligence is still in its infancy.
ARTIFICIAL INTELLIGENCE AND " MADE IN GERMANY”
Germany has an artificial intelligence research institution established in 1988. This institution is run by member companies. In addition, some departments of Fraunhofer Institutes and universities are also focused on this issue, but the leadership in artificial intelligence applications in the world belongs to China, especially the United States.
Germany has created a new strategy to maintain and improve competitiveness, not to fall further behind in artificial intelligence applications around the world. Under the theme” Artificial Intelligence-Made in Germany", the Federal Government announced a 3 billion euro Support Program and expects as much investment from the private sector. According to this program, effective from 2019;
• A support-motivation program for the creation of 100 new professorship quotas in universities only for this field and the withdrawal of Science and research people from abroad to Germany,
* Tax-exempt investments of enterprises in this area,
* Establishment of artificial intelligence business knowledge centers on a provincial or regional basis,
* It is envisaged to establish real laboratories where artificial intelligence applications can be tested.
At the heart of the German artificial intelligence strategy are machine learning, deep learning, machine-human interaction and self-learning systems.
In Turkey's digitalization strategy, planning should be done according to sectors or even sub-sector groups. Especially for export products, this work should be started immediately. In this planning, the three concepts I'm talking about have to be considered together and executed together. And we don't need to think about how this happens, there are countless experimental applications in front of us. We need to analyze them well and start taking quick steps on this issue as soon as possible.