Machine Learning-Stanford: Learning note 1-motivation and application of machine learning

Source: Internet
Author: User

The motive and application of machine learning

Tools: Need genuine: Matlab, free: Octave

definition (Arthur Samuel 1959):

The research field that gives the computer learning ability without directly programming the problem.

Example: Arthur's chess procedure, calculates the probability of winning each step, and eventually defeats the program author himself. (Feel the idea of using decision trees)

definition 2(Tom Mitchell 1998):

A reasonable learning problem should be defined as: For a computer program, give it a task T and a performance measurement method p, if under the influence of experience E, p to t measurement results have been improved, then said the change program from E learning.

As in the example above: E: The program constantly and their own experience in chess, T: Chess, P: and human players play a winning percentage

Four parts of the course:

1. Supervised learning

(1) Regression problems

Example: Collecting price statistics for a certain house

Housing size and price correspondence:

By drawing a fitting curve, you can estimate the price by the size of the house.

-supervised learning gives a data set (correct house price and corresponding size)

-This example is a regression problem. regression means that the variables that need to be predicted are continuous.

(2) Classification problems

The variables to be dealt with in the classification problem are discrete

Example: Judging whether the tumor is malignant or bisexual

-Collect tumor size and malignant/benign data, size horizontal axis, whether it is malignant for longitudinal axes (only 0, 1) draw

-the tumour may be caused by a number of factors, the introduction of age, the size of the horizontal axis, the age of longitudinal axes, malignant with a fork, benign with a circle to express paint, analysis of the area of the tumor

-You can also introduce more properties, drawn in multidimensional space

-How does an infinite dimensional space work? An algorithm that maps an infinite dimension to memory?

2. Learning theory

Learning theory explains the effective reason for learning-based algorithms (the theoretical basis of learning algorithms)

Find out what algorithms can approximate different functions well, and the size of the training set is appropriate.

3. No supervised learning

Example: In the case of the tumour above, the point in the figure does not know the correct answer, but is from you to find a certain structure, that is, clustering .

Applied in the fields of biological genetic engineering, image processing, computer vision, etc.

Example: Cocktail party issues

Pick up the sounds you're interested in during a noisy cocktail party

Use two different positions to separate the sound from different positions

can also be used in text processing and other fields

Using the ICA algorithm, MATLAB line of code can be solved

4. Intensive Learning

A series of decision-making is produced by the conclusion or right or wrong.

Example: to a model plane to write a takeoff program, the aircraft in the program made a series of wrong decisions is to crash, as long as the overall good decision to make a continuous, can keep the aircraft normal flight

The basic concepts of reinforcement learning: return function (positive feedback and negative feedback), positive feedback when the program makes the right decision, and vice versa.

The process is constantly making decisions, learning and making the right decisions as you try to get as much positive feedback as possible.

The key is to define what is the right decision , what is the wrong decision, and then design the algorithm to get as much positive feedback

Machine Learning-Stanford: Learning note 1-motivation and application of machine learning

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: info-contact@alibabacloud.com and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.