University of Jordan Artificial intelligence
Dr. Nadim Ali Meri Obeid
Project name: Machine Learning

Major Student number Name
CIS 0091421 Ahmad Fayed
CIS 0094617 Haytham Issam

Machine Learning
(The Discipline of Machine Learning 1st Paper)

Considering the things that I’ve read about the subject of machine learning; I think most of them tried to develop machine learning by trying to give the machine itself the answers to a vast variety of questions.
Also they seem to define 3 aspects in which the machine can learn with respect to, which are:
1. Performance (P).
2. Task (T).
3. Experience (E).

- My perspective on machine learning so far is that it’s the science of making computers able to conclude, suggest, and give answers to indirect questions and maybe even program themselves using initial structures, experience, and statistics.

- Machine learning scientists need to have some source of inspiration or some kind of a source to make machines learn like, so they study and analyze human and animal learning processes through psychology and neuroscience.

- I read some online articles besides the paper I’m summarizing, and I found that they all mentioned examples on the success of some machine learning based applications so I think I have to list a few.

1) Speech recognition: now these systems use machine learning. Because the speech recognition accuracy is greater if one trains the system, than if one attempts to program it by hand.

2) Robot control: in my opinion this is where machine learning really shines many machine learning strategies have been successfully applied to control robots. As an example they could control a stable helicopter flight.

- It is true that machine learning is one great method for developing applications, but one cannot just use it to develop any kind of software you have to consider the nature of the application.

- The machine learning method is applied best when the following criterias are found in the nature of the application:

1) If the application is too complicated and designing an algorithm is very hard, it’s best to use machine learning, even if the algorithm could be designed, it won’t give the same level of accuracy.

2) If the application requires adaptation to its changing environment after it’s applied. In this case, the flexibility and the dynamicity machine learning gives to the applications developed by it are known to be more effective and to have better performance.

Personal Note: when I got to the upper part i realized that maybe machine learning method can be as good as the nature of the application fits to be developed by it, so the more fitting the application is to be developed by machine learning the more effective machine learning method will be.

- In the research area of machine learning there are some main questions to search about some of them is how to be able to transfer what is learned in one task by the machine and make use of it in other related tasks?
- Another one is what is the relationship between learning algorithms and how to decide which one is to be used in a particular time.

- In everything related to AI ethical issues must come to the discussion. It’s a sure thing that technologies are developing rapidly and in the future machine learning will be applied vastly in every field so an ethical question comes and says “how can we decide when to apply these technologies?” and “ wouldn’t applying an X technology invade the personal privacy” etc…. Well Dr. Nadeem. I will make this part short. I think as long as the benefits of applying some technology are more than the side effects or disadvantages if you will, it should be applied as soon as it is ready

Introduction to Machine Learning for AI (2nd Paper)

- Intelligence

We can define the intelligence in many ways; in this context the intelligence is the ability to take the right decisions using many criterias requiring knowledge.

- Artificial Intelligence

Computer one of the things that already helps human in many ways that means the computer can take the right decisions, but animals and humans can do many things that computers can not reach basically because the scientists do not know how to do these things.

The most popular concepts in the subject

1) Pattern recognition.
2) Neutral networks.
3) Pattern classification.

- My opinion

The human can live without AI but the AI make the human life easier, but the human must not depend on the AI in everything because there is some tasks that the scientists cannot program it.