High Algorithm - Distribution of Algorithms by Levels of Difficulty.
Good time to read, dear users of Habr!
Pedro Domingos’s High Algorithm describes the families of various algorithms used in the design of artificial intelligence systems.
The proposed article provides arguments for the specialization of algorithms by levels of complexity.
For algorithms, Domingos offers a chain of five varieties of algorithms, each of which can evaluate a different algorithm at a certain stage of the study. This article assumes that the algorithms logically follow one after the other. To ensure such a sequence, a pair of algorithms (Bayesian and evolutionary) had to be swapped, and also to change the entry point to the family of algorithms. In the proposed model, the chain begins with processing using the Bayes theorem.
I will give the main sections (levels of complexity) that can be explored to improve the interaction of existing algorithms.
Supergroup - sections of the classification of applications:
Inanimate - the inanimate nature
Animate - living beings, including people
Cosmic - science, including artificial intelligence
Subgroup - AI algorithms, including
Bayes science - definition of random connections
Evolution - evolutionary algorithms
Analogy - pattern recognition
Symbolic - symbolic calculations
Gradient - mathematical optimization
Let's start with a level about which little is currently known - whether the strings fly there, or dark matter.
Chaos - Inanimate - Bayes - unrecognizable primary chaos The
laws of evolution are confirmed in the animal world, but evolutionary algorithms also apply to the microworld with the macrocosm. For example, the rotation of one object around another.
Fractal - Inanimate - Evolution - a fractal structure for cognition
By the energy according to Einstein's formula, both matter having mass and weightless photons are reduced.
Energy - Inanimate - Analogy - energy
Information properties largely outpace the appearance of a substance. Information includes not only entropy, but also such properties of the Universe as space and time.
Information- Inanimate - Symbolic - information
Systems with elements belong to the static level. There are a large number of artificial systems, but geology and astronomy are also involved in natural systems.
Static - Inanimate - Gradient - systems
Processes (mental, social, economic) distinguish wildlife from inanimate
Dynamic - Animate - Bayes - processes
Synergistic phenomena in which a person controls the elements of inanimate nature to change the habitat, allow you to organize the living space in accordance with the needs.
Market - Animate - Evolution - the market
The ability to unite "by interests" of the
Corporation - Animate - Analogy - corporation plays an important role in organizing large groups.
With further complication, the rules are fixed and specialized bodies appear to monitor the rules.
Bureaucratic - Animate - Symbolic - bureaucratic apparatus
When approaching the limit of the population there is a need to control activities that support existing environmental systems.
Ecology - Animate - Gradient - ecology
As humankind (or artificial beings) works, it becomes necessary to expand the occupied territory, which now leads to attempts to conquer outer space.
Space - Cosmic - Bayes - outer space The
exchange of ideas to expand knowledge is similar to the mechanism of animal evolution.
Intellect - Cosmic - Evolution - development of science
As the number of ideas increases, the need for their classification increases.
Class - Cosmic - Analogy - classification of phenomena.
After classification, it becomes possible to determine the reactions of class members to environmental changes.
General - Cosmic - Symbolic - definition of the laws of nature
It is possible to optimize living and non-living nature to favor the development of nature and society.
Optimal - Cosmic - Gradient - determination of the optimal structure.
The beginning of the next round of studying random manifestations is possible.
Development - preparation
Progress - development
Stabilization - stabilization
Conservation - conservation (not used on its own, disguised as Development of the next level of development)
Matter - material
Abstract - abstract
It is assumed that this set of attributes constitutes a necessary subset for applying known machine learning algorithms. Using the table of levels of complexity allows you to test the hypothesis of supergroups, subgroups, levels of complexity on data posted on the Internet.
Links:
habr.com/post/259291 - Tunnel modeling - version 0.9
habr.com/en/post/316198 - Proposal for modifying the rules of the game Life
Pedro Domingos’s High Algorithm describes the families of various algorithms used in the design of artificial intelligence systems.
The proposed article provides arguments for the specialization of algorithms by levels of complexity.
For algorithms, Domingos offers a chain of five varieties of algorithms, each of which can evaluate a different algorithm at a certain stage of the study. This article assumes that the algorithms logically follow one after the other. To ensure such a sequence, a pair of algorithms (Bayesian and evolutionary) had to be swapped, and also to change the entry point to the family of algorithms. In the proposed model, the chain begins with processing using the Bayes theorem.
I will give the main sections (levels of complexity) that can be explored to improve the interaction of existing algorithms.
Supergroup - sections of the classification of applications:
Inanimate - the inanimate nature
Animate - living beings, including people
Cosmic - science, including artificial intelligence
Subgroup - AI algorithms, including
Bayes science - definition of random connections
Evolution - evolutionary algorithms
Analogy - pattern recognition
Symbolic - symbolic calculations
Gradient - mathematical optimization
Level - individual difficulty levels
Inanimate nature:
Let's start with a level about which little is currently known - whether the strings fly there, or dark matter.
Chaos - Inanimate - Bayes - unrecognizable primary chaos The
laws of evolution are confirmed in the animal world, but evolutionary algorithms also apply to the microworld with the macrocosm. For example, the rotation of one object around another.
Fractal - Inanimate - Evolution - a fractal structure for cognition
By the energy according to Einstein's formula, both matter having mass and weightless photons are reduced.
Energy - Inanimate - Analogy - energy
Information properties largely outpace the appearance of a substance. Information includes not only entropy, but also such properties of the Universe as space and time.
Information- Inanimate - Symbolic - information
Systems with elements belong to the static level. There are a large number of artificial systems, but geology and astronomy are also involved in natural systems.
Static - Inanimate - Gradient - systems
Nature:
Processes (mental, social, economic) distinguish wildlife from inanimate
Dynamic - Animate - Bayes - processes
Synergistic phenomena in which a person controls the elements of inanimate nature to change the habitat, allow you to organize the living space in accordance with the needs.
Market - Animate - Evolution - the market
The ability to unite "by interests" of the
Corporation - Animate - Analogy - corporation plays an important role in organizing large groups.
With further complication, the rules are fixed and specialized bodies appear to monitor the rules.
Bureaucratic - Animate - Symbolic - bureaucratic apparatus
When approaching the limit of the population there is a need to control activities that support existing environmental systems.
Ecology - Animate - Gradient - ecology
superhuman decisions:
As humankind (or artificial beings) works, it becomes necessary to expand the occupied territory, which now leads to attempts to conquer outer space.
Space - Cosmic - Bayes - outer space The
exchange of ideas to expand knowledge is similar to the mechanism of animal evolution.
Intellect - Cosmic - Evolution - development of science
As the number of ideas increases, the need for their classification increases.
Class - Cosmic - Analogy - classification of phenomena.
After classification, it becomes possible to determine the reactions of class members to environmental changes.
General - Cosmic - Symbolic - definition of the laws of nature
It is possible to optimize living and non-living nature to favor the development of nature and society.
Optimal - Cosmic - Gradient - determination of the optimal structure.
The beginning of the next round of studying random manifestations is possible.
Periods of developmental levels:
Development - preparation
Progress - development
Stabilization - stabilization
Conservation - conservation (not used on its own, disguised as Development of the next level of development)
Abstraction - levels of abstractness
Matter - material
Abstract - abstract
Conclusion:
It is assumed that this set of attributes constitutes a necessary subset for applying known machine learning algorithms. Using the table of levels of complexity allows you to test the hypothesis of supergroups, subgroups, levels of complexity on data posted on the Internet.
Links:
habr.com/post/259291 - Tunnel modeling - version 0.9
habr.com/en/post/316198 - Proposal for modifying the rules of the game Life