What does Artificial Intelligence mean and when did it start?
The word “intelligent” as defined in the dictionary, is the ability to acquire and apply knowledge and skills, through explaining a certain phenomenon relying on experience, acquired knowledge or expectation. Human intelligence can be evaluated through analyzing one’s decisions to several situations, depending on one’s experience, acquired knowledge and expectations of what might occur; and one’s ability to learn from previous mistakes in order to avoid its repetition in the future, and make better decisions when one experiences the same situation even with little changes in its circumstances.
So we started to direct our thoughts to machines, can they think? Are they capable of taking right decisions like humans?
Before we could answer this, it was first required to understand how the human brain works and how it learns from one’s mistakes. This was the first stage in the history of Artificial Intelligence that is considered as one of the modern sciences compared to other sciences.
In 1956, the two scientists, Allen Newell and Simon Herbert, developed the theory of ‘Logic Theorist’, LT for short. They tested it at a computer program when was the theory was formalized t, so they were capable of predicting expected results by the help of each other.
However, an unexpected result has occurred, were the theory failed, according to the program, because of only one parameter. This was because humans did not use the same strategy to solve a certain problem like how the program did to solve the same problem. Therefore, the theory was revised in order to reflect how humans reacted to solving problems, and was then named ‘GPS’ (General Problem Solver). The revised theory gave much better results, and was thus named ‘Ends Analysis’, and is now recognized as one of the strategies to solve problems in Artificial Intelligence.
In year 1950, a group of American psychologists, studied and understood psychology to improve data processing in developing AI (Information-Processing Psychology).It was considered as one of the essential and most challenging step to understand the human brain. This was because understanding and responding to situations vary from one person to another, and not only that, but also in analyzing them. In addition, to the understanding of the spoken words according to their understanding of the word itself.
Artificial Intelligence in currency and trade
Artificial Intelligence is not just a thought or a science fiction movie. It has tenanted wide fields, and has become a main refuge in taking crucial decisions for companies, institutions and some countries. These decisions are, without doubt, the most challenging that many managers, economic researchers, and workers in trade in general, might encounter. Such decisions were considered a substantial risk to their institutions, until it has become less risky, to a great extent, with the advances in science and AI programs.
Generally, a ‘knowledge engineer’ develops such programs where it is not necessary for him to have a background about a specific problem, but does enough with having full knowledge of financial sciences and theories used in economy; so that he’s able to place it within the program that can analyze inputs to output results closer to what can be accepted or considered along with using the help of financial experts and business decision-makers to make use of their expertise within the program.
Sometimes, it’s referred to as ‘ Rules-Based-Systems’ and is used in detecting metal, prospecting, medical diagnosis and fraud detection.
However, these systems were not able to set themselves free from all restrictions to make guaranteed decisions without having all information required to get a result, and it is difficult to have all factors within one program along with having many other variables, so it is still considered indefinite compared to the next.
Artificial Neural Networks in Finance
It is one of the fundamental pillars of Artificial Intelligence, which is used in some daily life applications like voice recognition and autonomous vehicles.
In the field of Finance, this technology is capable of dealing with uncertain data, to recognize some patterns and use these patterns to predict forthcoming events. Predicting results related to economic events, like change in interest rate and currency movements, one of the most challenging obstacles in the economy field and is considered as one of the condemnatory operations that requires an error of zero percent in short time.
So, it has proved to solve such problems when compared to other technologies, for its ability to handle a hell lot of complicated and changing data. Not only this, but also such algorithms are capable of recognizing mistakes and putting them into consideration to avoid repeating it unlike the ‘expert system’.
What are the tasks that ANNS solve?
Medsker incorporated the following financial functions in 1991
- Credit authorization screening
- Mortgage risk assessment
- Project management and bidding strategy
- Financial and economic forecasting
- Risk rating of exchange-traded, fixed income investments.
- Detection of regularities in security price movements
- Prediction of default and bankruptcy
Then, Hsieh added many others in year 1993, which included:
- Financial Simulation
- Predicting Investor’s Behavior
- Credit Approval
- Security and/or Asset Portfolio Management
- Pricing Initial Public Offerings
- Determining Optimal Capital Structure
How ANNs (Artificial Neural Networks) work?
ANNs models are inspired by biological sciences that study how humans use the neuroanatomy in solving problems. Dr. Robert Hecht Nielsen who is the inventor of first neurocomputers defined the ANNs as “a computing system made up of a number of simple, highly interconnected processing elements, which process information by their dynamic state response to external inputs. ANNs behave like humans, learn by observing examples. Learning in human is being processed by the synaptic connections that exit between neurons exactly the same approach in ANNs in Artificial manner.
In human brain, signals is being collected by neuron from others through dendrites, neuron sends out electrical activity spikes through a fine nerves called axon. This axon splits into a thousands of branches , at the end of each branch a structure called synapse which evaluates the electrical activity came from the axon either it is excite or inhibit activity in the connected neurons. In the end learning process is accomplished by changing in the synapses effectiveness so that the impact of neuron on another changes.
By Idealization what happens in real human’s neurons, typically features are simulated in artificial neurons by different approaches ended up simple neurons to the complicated neurons.
Learning process in ANNs
We can classify learning process in two main categories
- Supervised Learning: it depends on an existing of an external teacher that tells output what is the desired response should be according to the given input signals. Paradigm examples of supervised learning include error-correction learning and stochastic learning.
- Unsupervised Learning: it is not teacher dependent but based only on local information. It self organizes data in the network and detect its properties. Usually unsupervised learning process works off-line which means there is a separation between the two learning and operation face unlike the supervised learning process which learns on-line.
Artificial intelligence has not proven its importance in smart vehicles and machines only but also in taking critical decisions in business, Medicine and all disciplines AI maybe threatens some others current jobs and takes over our career lives.