Machine learning has the wide spectrum of applications including search engines, medical diagnosis, detecting credit card fraud, stock market analysis, classifying DNA sequences, speech and handwriting recognition, game playing and robot locomotion.

Human interaction

the bit of machine learning systems attempt to eliminate a want for person intuition in the analysis of the information, when others adopt a collaborative approach between man & machine. Human being intuition can't exist as totally eliminated since the designer of a models must specify how else the information come to exist as represented & what mechanisms is utilized to lookup for a characterization of the information. Machine learning may be hold an attempt to automate area of the scientific method. Occasionally machine learning research worker produce methods inside a framework of Bayesian statistics.

Algorithm types

Machine learning algorithms are organized into the taxonomy, based on a desired effect of the algorithmic rule. Most common algorithmic program types include:

supervised learning --- where the algorithmic rule generates a work that maps inputs to desired outputs. 1 standard formulation of the supervised learning project is the classification problem: the learner is expected to view (to approximate the behavior of) the work which maps a vector $[X\_1,\; X\_2,\; \backslash ldots\; X\_N]$ into one of many classes by searching at many input-output examples of the work. unsupervised learning --- which models the placed of inputs: tagged examples are non available. semi-supervised learning --- which combines both labelled & unlabelled examples to generate an appropriate work or even classifier. reinforcement learning --- where the algorithmic program learns a policy of training work given an observation of the globe. Each action has a select few impact in a epa, & the environment will bring feedback that guides the learning algorithmic rule. transduction --- similar to supervised learning, but doesn't explicitly construct the work: instead, endeavor to predict newly outputs according to step by step step by step videos inputs, training outputs, & fresh inputs. learning to learn --- where a algorithmic rule learns its have inductive bias based on previous case.

A performance & computational analysis of machine learning algorithmic rule occurs as branch of statistics known as computational learning theory.

Machine learning topics

This names is the topics covered in a average machine learning course.

Modeling conditional probability density functions: regression and classification Artificial neural networks Decision trees Gene expression programming Genetic Programming Gaussian process regression Linear discriminant analysis k-Nearest Neighbor Minimum message length Perceptron Radial basis functions Support vector machines Modeling probability density functions through generative models: Expectation-maximization algorithm Graphical models including Bayesian networks and Markov Random Fields Generative Topographic Mapping Appromixate illation techniques: Markov chain Monte Carlo method Variational methods Optimization: most of methods enrolled above either utilize optimisation or even come cases of optimisation algorithmic program.