Artificial Intelligence and Machine Learning Fundamentals

Artificial Intelligence and Machine Learning Fundamentals.


In the series of artificial intelligence, machine learning and evolutionary computing this is my first post. this post describes motivation and basics behind artificial intelligence in a simple manner. In upcoming posts I will elaborate concepts of artificial intelligence, its branches i.e. Artificial Neural networks, Fuzzy Systems  and evolutionary algorithms, related applications, algorithms with some code samples and benchmark tests. So lets first see what artificial intelligence is all about.
Literal definition of Artificial Intelligence is as follows
“Artificial Intelligence is an area of computer science concerned with designing intelligent computer systems, that is, systems that exhibit the characteristics we associate with intelligence in human behavior. In other words AI is a science that is concerned with automation of intelligent behavior“
However the term intelligence is not very well defined and therefore has been less understood. Consequently, task associated with intelligence such as learning, intuition, creativity, and inference all seem to have been partially understood. Artificial Intelligence van be divided into two branches
  1. Strong AI: strong AI exists only in hypothesis. Believers of strong AI propose that strong AI machines can perform task better or at-least comparable to their human counterpart. Numerous movies (AI , I,Robot, Terminator series) and countless science fiction novels advocate the idea of super intelligent humanoids and machines that can think, do Intellectual stuff and  poses threat to humanity are influenced by the idea of strong AI, but believe me these things are not going to happen in near future. one of the biggest dilemma in strong AI is to outperform itself, it means that the moment you understand how a particular AI mechanism works it is  possible to  beat it, at-least in theory.
  2. Weak AI: Weak AI or application oriented AI tries to solve real world problems by taking inspiration from Nature. Intelligence can be found in nature in various forms such as processing of human brain, human problem solving capabilities, process of natural evolution, flock of birds searching for food, ant optimizing path from food source to ant colony and many more. in weak AI we try to implement these behavioral intelligence into computer systems. Some of the weak AI technologies include but are not limited to expert systems, neural networks, fuzzy logic, cellular automata, and probabilistic reasoning. Out of these technologies, neural networks, fuzzy logic and probabilistic reasoning are predominantly known as “soft computing”. The term soft computing was introduced by Lotfi A. Zadeh of the University of California, Berkley, U.S.A. Probabilistic reasoning subsumes genetic algorithms, chaos, swarm optimization, simulated annealing etc. and part of learning theory. In my future references or posts i will use the terms soft computing, Artificial Intelligence and machine learning interchangeably in context of Weak AI. Lets leave the topic of Strong AI to fiction writers and filmmakers.
Now the question arises what is the need of AI based systems when we have the perfect world of mathematical equations and models in place. To appreciate the usefulness of AI Based systems first we need to understand pitfalls of mathematical models.
Traditional modeling of physical processes is often named as white box modeling or physically-based modeling (or knowledge-driven modeling) because it tries to explain the underlying processes.A white box model or a clear glass model is a system where all necessary information is available to model the system. A black-box model is a system of which there is no a priori information is available and majority of soft computing methods lie in the zone of either black-box or grey box (hybrid of white box and black box) modeling. Usually it is preferable to use as much a priori information as possible to make the model more accurate, therefore white box models are considered easier provided use of a priori information is accurate.
As the complexity of the system increases white box modeling or hard computingbecomes more and more tedious, at a certain degree of complexity it is nearly impossible to model a system accurately, here comes black box modeling or soft computing into picture with a virtue of no need of a priori information, its fault tolerance capabilities and its capability of generalization.
Black box modeling or data-driven models or soft computing borrowing heavily from Artificial Intelligence (AI) techniques, are based on a limited knowledge of the modeling process and rely on the data describing input and output characteristics. These methods, however, are able to make abstractions and generalizations of the process and play often a complementary role to physically-based models. Data-driven modeling uses results from such overlapping fields as data mining, artificial neural networks (ANN), rule-based type approaches such as expert systems, fuzzy logic concepts, rule-induction and machine learning systems. Sometimes “hybrid models” are built combining both types of models.
A simple example of a data-driven model is a linear regression model. Coefficients of the regression equation are identified (“trained”) on the basis of the available existing data. Then for a given new value of the independent (input) variable it gives an approximation of an output variable value. More complex data-driven models are highly non-linear, allowing many inputs and many outputs they need a considerable amount of historical data to be trained, and if this is done properly, they are able not only to approximate practically any given function, but also to generalize, providing correct output for the previously “unseen” inputs.
The question “Which model?” has been asked ever since modeling started, but it still brings about a lot of confusion and disagreement. Numerous models have been and are constantly being proposed, almost each one of which claiming to add something to the knowledge of a phenomenon or process, but each inherently inadequate to encompass the full complexity of the real world. Comparison can be drawn between two methods of modeling (white box and black box):
White box, hard or physical modeling
Black box, soft or empirical modeling
Mathematical approach
Biological systems
Less precise
Fault tolerance
A priori  knowledge
Not required
Complexity handling capabilities
predictive potential
Theoretically good but decreases rapidly with increase of complexity
Theoretically poor but sustainable at higher degree of complexity
Less than soft computing
Training data set requirements
Not needed but inputs must be highly precise
Handling inverse problems
In reality, both empirical and physical models have found various and successful implementations and developments in different scientific areas. But as the complexity of the process increases, which is the general case, advantages of physical models rapidly diminishes and data driven models increases and at high complexity where there is no explicit relation between input and output parameters the superiority of data driven model is unmatched.
in the next post I will be discussing about various branches of Artificial Intelligence and  their applications in real world problems.
kindly post your comments in the comment section. for any queries you can mail me at

2 thoughts on “Artificial Intelligence and Machine Learning Fundamentals

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s