Introduction

Machine Learning

Def 1 : 

The field of study that gives computers the ability to learn without being explicitly programmed.     --Arthur Samuel

Def 2 :

 A computer program is said to learn from experience E with respect to some class  of tasks T and performance measured P, if its performance at tasks in T, as measured by P, improve with experience E.     --Tom Mitchell

Any ML problems can assigned to one of two broad classifications:

• Supervised learning

    We are given a data set and already know what our correct output should look like, having the idea that there is a relationship between the input and the output.  
 
In a regression problem, we are trying to predict results within a continuous output.
In a classification problem, we are instead trying to predict results in a discrete output.

Example :

   
     Regression - 
    Given a picture of a person, we have to predict their age on the basis of the given picture

    Classification - 

    Given a patient with a tumor, we have to predict whether the tumor is malignant or benign.

• Unsupervised learning

    Unsupervised learning allows us to approach problems with little or no idea what our results should look like. We can derive structure from data where we don't necessarily know the effect of the variables.

Example :

    

    Clustering-

    Take a collection of 1,000 different genes, and find a way to group these genes into groups that are similar or related by different variables.

    Non-Clustering-

    The "Cocktail Party Algorithm", this algorithm can identify individual voices and music from a mesh of sounds at a cocktail party.

At last, all the article is the course's note from the "coursera", when it added the label [Machine Learning teached by Andrew Ng] 

Fuzzy Logic brief introduction

Fuzzy Logic

In the traditional computer logic, the truth value could be 0 or 1, so if we ask a question to computer :

" Is it cold today? " the answer will only "Yes" or "No". It cannot understand " a little cold" or " very

cold ", but Fuzzy logic can handle the concept of truth with many value between completely true and

completely false, it helps AI to take an important step.

As the below plot, the temperature is separated by dichotomy logic, cold or not called crisp set.

 But human's feeling often contain ambiguous narratives in semantics, in order to express more

accurately, the dichotomy extend to many-valued logic.

An fuzzy set called "membership function", it is a set of numbers which mapping between truth and

psychological. For example,  the following are 4 commonly used membership functions:

We can understand the relationship easily between the temperature and the degree of each word.

Fuzzy Inference

rules:

IF ___ IS ___ , THEN ___ IS ___ .

ex:

IF _temperature_ IS _ cold_ , THEN _fan speed_ IS _slow_ .

IF _temperature_ IS _ hot_ , THEN _fan speed_ IS _high_ .

Zadeh operators

NOT x = (1 - truth(x))

x AND y = minimum(truth(x), truth(y))

x OR y = maximum(truth(x), truth(y))

Fuzzy Logic vs Probability

Fuzzy logic is using pretty much the same tools as probability theory. But it's using them to trying to

capture a very different idea. Fuzzy logic is all about degrees of truth - about fuzziness and partial or

relative truths. Probability theory is interested in trying to make predictions about events from a state

of partial knowledge. But probability theory says nothing about how to reason about things that aren't

entirely true or false.