Machine Learning

-

What is Machine Learning ?

Machine Learning is concerned with design and development of algorithms which allows a computer to improve performance  based on data.

Applications of Machine Learning
  • Gmail - Priority inbox  :A sample of user defined important mails are given and machine should identify new important  mails.   
  • Handwritten digit recognition
  • Generating captions with sentiments.
  • Autonomous robotics
  • bioinformatics
  • Medical diagnosis
  • Stock market analysis

Types of Machine Learning
  • Supervised Learning       : A pair of a training data and a target is given at a time. At each step,                                            machine learns and refine a mapping form
  • Unsupervised Learning   : Only the data is given. Machine learns the structure inherited in data.
  • Reinforcement Learning : Used in game playing.
  • Other Machine Learning Types 
  1.  Active Learning :   When building a model, algorithm may choose which data to come next at                                   runtime.
  2.  Transduction :  When building a model, algorithm is allowed to use the test data.
  3.  Co-training with two different but related data sets
  4.  Estimation with missing variables               

Training Regimes
  • Batch leaning
   All training data and targets are given. The machine learns how to map training data to
   targets which can then be applied to yet unseen data.
  • Online Processing
  A pair of a training data and a target is given at a time. At each step, Machine learns and refines a 
  mapping form which can then be applied to yet unseen data.             


Comments

Post a Comment

Popular posts from this blog

Normalization

-
Image
Normalization is a process for evaluating and correcting table structures to minimize data redundancies, thereby reducing the likelihood of data anomalies. Normalization starts by identifying the dependencies of a given relation and progressively breaking up the relation (table) into a set of new relations (tables) based on the identified dependencies. Normal Forms 0NF            Multi-valued attributes exists 1NF           Any multi-valued attributes (repeating groups) have been removed. 2NF           Any partial functional dependencies have been removed 3NF           Any transitive dependencies have been removed BCNF           Any anomalies resulting from functional dependencies have been removed 4NF           Any multi valued dependencies have been removed 5NF           Any remaining anomalies have been removed The table in 0NF : eliminate the nulls by making sure that each repeating group attribute contains an appropriate data va
Read more

Database Types

-
Types of Databases Databases can be classified according to the number of users, o    Single user database - A single-user database that runs on a personal computer is called a desktop database o    Multi user database ·          Workgroup database - the multiuser database supports a relatively small number of users (usually fewer than 50) or a specific department within an organization ·          Enterprise database Databases can be classified according to the database location(s), o    Centralized database - a database that supports data located at a single Site. o    Distributed database – a database that supports data distributed across several different sites The most popular way of classifying databases today, however, is based on how they will be used and on the time sensitivity of the information gathered from them. o    operational database (sometimes referred to as a transactional or production database) - A database that is designed primarily to suppor
Read more

Entity Relationship Diagram (Pat II)

-
Image
Mapping Binary Relationships One to one Relationship  Partial participation (Partial participation is represented by single lines.)           Not every professor teaches every student. Every student is not learnt by every professor.             Relational Mapping                                                                   Professor ( Name , Phone , Address )                                                                                       ↑FK            Course   ( CourseId  , CourseName , Credits , Name) Total Participation (Total participation is represented by double lines.)             Every professor teaches every student. Every student is learnt by every professor           Relational Mapping                      Professor ( Name , Phone , Address , CourseId , CourseName , Credits) One to many Relationship (Partial participation)           Relational Mapping                                          
Read more