Machine learning - machine learning technology - Electrical input

Thursday, April 18, 2019

Machine learning - machine learning technology





Machine learning is the study of statical models and algorithms that computers use to productively complete a particular task without using direct instruction, relaying on ornament and conclusion instead. Machine learning generate mathematical model based on illustrative data which is also known as training data, orderly make decision and prediction without being directly programmed to complete the task.

Applications for machine learning:-

1       1.      Agriculture
2       2.      Computer vision
3       3.    Computer Networks
4       4.       Bioinformatics
5       5.       Adaptive websites
6       6.       Anatomy
7       7.       Financial market analysis
8       8.       DNA sequence classification
9       9.       Machine learning control
1      10.     Medical diagnosis
1      11.     Telecommunication
        12.   General game playing
1      13.   Data quality
1      14.   Search engines
        15.   Optimization
        16.   User behavior analytics
        17.   Machine perception
        18.   Internet fraud detection
        19.   Anatomy
        20.   Brain-machine interfaces

Different types of learning algorithms

Ø  Supervised learning
Ø  Semi- supervised learning
Ø  Unsupervised learning
Ø  Reinforcement learning

Supervised learning:-
is the machine learning task that maps example  input and output pairs. In supervisor learning each pair of example be composed of an input object(vector) and a desired output which is also called supervisory signal. Supervisory learning algorithm examine training data and produce an deduce function, which can be used for mapping new example. In semi-supervised some of training example are missing the desired output.

Most widely used learning algorithms are:-
1.    Naive Bayes
2.    Decision trees
3.    K-nearest neighbor algorithm
4.    Linear regression
5.    Similarity learning
6.    Neural Networks
7.    Linear discriminant analysis

Unsupervised learning:-
is a type of machine learning which groups data that has not been labelled. Unsupervised algorithm used to draw intrusions data files which consisting of input data without labeled responses. With the help of cluster analysis a data analysis to find hidden pattern. Clustering or cluster analysis grouping a set of objects where objects in the same group.

Most common unsupervised learning algorithm:-

          1.    Clustering
          2.    k- means
          3.    OPTICS algorithm
          4.    Mixture models
          5.    Anomaly detection
          6.    Neural networks
          7.    Hebbian learning
          8.    Autoencoders
          9.    DBSCAN
         10.  Hierachical clustering

Reinforcement learning algorithm :-

It is the area of machine learning in which suitable action is taking to maximize some notion of progressive reward. Three basic concept in reinforcement learning state, action, reward. The state determine the current situation.

Application of reinforcement:-

         1.    Computer games
         2.    Online advertising
         3.    Robot control
         4.    Dialogue generation

No comments:

Post a Comment