We write you from Alliance4ai SUP’COM Tunisia to introduce you to the exciting world of artificial intelligence.
Recently there has been a lot of fuss about the word AI and we always heard about the definition of Artificial intelligence (AI) as the ability of a machine or computer system to copy human intelligence processes, learn from experiences, adapt to new information and perform human-like activities.
But to many of us, the idea is still unclear, confusing and many questions stand in our heads: How it works ? Is it a black box? and how scientists teach computers to achieve human-like tasks or in some cases, even outperform humans?
To make things clearer we have tried to answer to you these 2 basic questions.
1-Why machines need Learning ?
Before learning took place, problem-solving tasks relied on writing algorithms. An algorithm is simply a set of instructions that takes an input and returns an output as a solution for the problem.
Consider the following: Given a list of numbers, you are asked to sort them in an increasing order. This problem is solved by algorithms.
On the other hand, some problems were not so easy to solve by algorithms. People started to ask more from computers. They wanted the machine to have super abilities of solving very hard tasks. Tasks that scientists completely had no idea how to program.
For example: How is it possible to write an algorithm, that takes an image of an animal and outputs the type of it? This is a very easy task for humans, but solving it with algorithms is a very complex mission if not impossible. Humans know how to classify animal photos, but they do not know how to describe the steps they take to reach the answer. Here an important question arises. How do human learn ?
2- how do humans learn ?
Learning has been defined as a cumulative process of acquiring information and developing skills to figure out how to solve new problems and deal with new situations.
There are multiple techniques for teaching humans how to learn. They provide different ways of developing algorithms in the brain of how data patterns correspond to the outcomes. We tried to illustrate three learning techniques, and show how humans and machines learn using them.
2.1. Directed Learning
This is a simple method of providing exact step-by-step instructions to perform a given task.This is typically used for training workers on traditional assembly lines to execute a specific task. The only learning that happens is on how to exactly follow a procedure. There is no learning on dealing with the situation for which no instructions are provided
This is how most of the classic computer programs work. A program is essentially a sequence of steps that a computer must blindly execute — exactly as designed by a human programmer. Crashes occur when unplanned situations happen because the computer has no intelligence to deal with such a situation. Earlier AI systems were victims of this approach since human programmers had essentially codified their interpretation of all possible data patterns in an IF-THEN-ELSE logic. The AI machine itself was not capable of drawing it down conclusions. It had never really learned to deal with an un-programmed situation.
2.2. Assisted Learning
This is how a good teacher teaches students. The focus of teaching is not just to learn the exact subject matter, but also to learn how to understand the generic topic.
Teaching is done by examples. This helps the student understand the data patterns and their outcome in the examples used. With more and more practice, the student can deal with most situations — however, ambiguous or unpredictable — by using the self-developed algorithm.
This is also how IBM Watson works. To make Watson the best expert at diagnosing cancer, it is fed with millions of past cases of cancer with details on symptoms, recommended treatment and outcomes. These cases provide Watson with patterns of symptoms and treatments leading to outcomes. Neural networks within Watson develop its generic algorithms based on the patterns in this massive amount of information. It learns to diagnose cancer and recommend treatments. Google uses a similar technique for automatic translation; Facebook for face recognition; Amazon for generating product recommendation.
2.3. Self-Learning
We learn a lot ourselves, without a teacher. By repeatedly observing an activity we figure out the correlation between what someone does and its outcome.
Machines can learn in the same way. For example, if video streams from all surveillance cameras in a public place are fed into a Machine Learning AI system, over time the system is able to establish by itself the pattern of video data for a normal situation.
Conclusion
As we see, similar learning techniques are involved for humans and for machines. Both Learn via exposure to data patterns. Learning involves formulating an algorithm based on understanding data patterns and the resulting outcomes. Good learning requires exposure to a vast amount of data. A human or a system that has learned well is able to reliably predict the outcome. Better and faster predictions help to make better decisions. Better Decisions are critical for success !
About the Author
Alliance 4AI SUP’COM chapter is a part of the Alliance 4AI network implemented in the Higher School of Communication of Tunis – Tunisia (SUP’COM) one of the best Engineering Schools in Tunisia. The group of students who’ve written this article are all members of the Alliance 4AI community who share the same passion and desire to learn AI and create a global impact through its power and applications.
Links:
How AI machines learn-just like humans:
https://magazine.vunela.com/how-humans-and-machines-learn-c48de5360527
AI-Artificial-Intelligence:
https://searchenterpriseai.techtarget.com/definition/AI-Artificial-Intelligence