FAIF - Fast(Funny) Artificial Intelligence Framework
----------------------------------------------------
The C++ header only library for bioinformatics and artificial intelligence.
This library defines basic abstractions (like nomnial value, domain, point, space, state, etc.)
and implements classifier algorithms (Naive Bayesian, Decision Tree ID3 inspired, K Nearest Neighbors);
cross validator;
space search methods ( Evolutionary Algorithm, Expectaiton-Maximization, Hill Climbing);
timeseries primitives (TimeSeriesDigit, TimeSeriesReal, linear resampling, autoregressive predictor);
DNA primitives (Nucleotide, Chain, EnergyNucleo, SecStruct, FoldedChain, FoldedPair, Codon, CodonAminoTable,
Nussinov algorithm); Random generators; Gaussian eliminator.
Serialization to text or XML based on boost::serialization.
The FAIF abstractions and algorithms are generic (as STL and boost).
The main goal is education of AI, where simple and portable C++ library is needed.
The library will be used also for broad spectrum of applications
where some artificial intelligence algorithm are required.
Implementation:
--------------
C++ ISO 2003,
Dependencies:
-----------
boost 1.35 or later
Changelog
---------
0.44
+ parameters for decision tree and random forest classifiers
0.43
+ speed improvement in svm
0.42
+ faif/random used in svm
0.41
+ getCategory and getCategories for SVM classifier
0.40
+ polynomial and hyperbolic tangent kernels added to SVM classifier
0.39
+ support vector machine classifier is template (Val is template parameter)
---------
0.38
+ support vector machine classifier using smo(sequential minimal optimization) algorithm
---------
0.37
+ random forest classifier
+ more classifier tests
0.36
----
+ c++11 compilation mode
+ boost 1.57
+ discretizer serialization
0.35
----
+ remove active object to mt4cpp.sourceforge.net
+ remove min_val and max_val tempates, use std::min and std::max
0.34
----
+ learning, new classifier - k nearest neighbors classifier
+ learning refactoring - all classifiers are templates, add Point, Space, Belief classes
0.33
----
+ learning, decision tree classifier (ID3)
+ categories with belief in tree node
+ storing state for naive bayesian classifier and decision tree classifier using boost::serializable
0.32
----
+ discretizer (conversion from double to int) added to timeseries folder
+ learning - create example for empty collection
+ timeseries - description of epsilon calculation
0.31
----
+ header only library
+ Naive Bayesian Classifier refactoring
+ NBCInternalState class to store the internal probabilities for storing/loading Naive Bayesian Classifier
0.30
----
0.29
----
+ description of random generators (beginning)
+ random generators - examples
+ random custom distribution - implementation and test
+ build with boost1.42 and msvc10
0.28
----
+ scheduler.executeAsynchronouslyAndWait work properly when command is INTERRUPTED
+ improved Evolutionary Algorithm: crossover policy, roulette wheel selection
+ compilation with boost 1.41
0.27
----
+ compilation with boost 1.40
+ progress notification of given command take into account progress recalculation
0.26
----
+ hill climbing algorithm (local search)
+ evolutionary algorithm (improvements, stop condition, chromosome as vector of bools)
0.25
----
+ compilation with boost 1.39
+ added the int_power template function (for fast calculate power of integer exponent)
