-- Hoogle documentation, generated by Haddock
-- See Hoogle, http://www.haskell.org/hoogle/


-- | Machine Learning Toolbox
--   
--   Haskell Machine Learning Toolkit includes various methods of
--   supervised learning: linear regression, logistic regression, SVN,
--   neural networks, etc. as well as some methods of unsupervised methods:
--   K-Means and PCA.
@package mltool
@version 0.2.0.0


-- | Common type definitions used in all modules.
module MachineLearning.Types

-- | Scalar Type (Double)
type R = R

-- | Vector Types (of Doubles)
type Vector = Vector R

-- | Matrix Types (of Doubles)
type Matrix = Matrix R


-- | Regularization.
module MachineLearning.Regularization

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization

-- | Calculates regularization for Model.cost function. It takes
--   regularization parameter and theta.
costReg :: Regularization -> Vector -> R

-- | Calculates regularization for Model.gradient function. It takes
--   regularization parameter and theta.
gradientReg :: Regularization -> Vector -> Vector


-- | Randon generation uitility functions.
module MachineLearning.Random

-- | Samples <tt>n</tt> (given as a second parameter) values from
--   <tt>list</tt> (given as a third parameter).
sample :: RandomGen g => g -> Int -> Vector a -> (Vector a, g)

-- | Samples <tt>n</tt> (given as a second parameter) values from
--   <tt>list</tt> (given as a third parameter) inside RandomMonad.
sampleM :: RandomGen g => Int -> Vector a -> Rand g (Vector a)

-- | Returns a list of random values distributed in a closed interval
--   <tt>range</tt>
getRandomRListM :: (RandomGen g, Random a) => Int -> (a, a) -> Rand g [a]

-- | Returns a vector of random values distributed in a closed interval
--   <tt>range</tt>
getRandomRVectorM :: RandomGen g => Int -> (R, R) -> Rand g Vector

-- | Returns a matrix of random values distributed in a closed interval
--   <tt>range</tt>
getRandomRMatrixM :: RandomGen g => Int -> Int -> (R, R) -> Rand g Matrix


-- | Various Weight Initialization algorithms.
module MachineLearning.NeuralNetwork.WeightInitialization

-- | Weight Initialization Algorithm discussed in Nguyen at al.:
--   <a>https://web.stanford.edu/class/ee373b/nninitialization.pdf</a>
--   Nguyen, Derrick, Widrow, B. Improving the learning speed of 2-layer
--   neural networks by choosing initial values of adaptive weights. In
--   Proc. IJCNN, 1990; 3: 21-26.
nguyen :: RandomGen g => (Int, Int) -> Rand g (Matrix, Matrix)

-- | Weight Initialization Algorithm discussed in He at al.:
--   <a>https://arxiv.org/abs/1502.01852</a> Kaiming He, Xiangyu Zhang,
--   Shaoqing Ren, Jian Sun. Delving Deep into Rectifiers: Surpassing
--   Human-Level Performance on ImageNet Classification.
he :: RandomGen g => (Int, Int) -> Rand g (Matrix, Matrix)


-- | Tanh Activation.
module MachineLearning.NeuralNetwork.TanhActivation
tanh :: Floating a => a -> a
gradient :: Matrix -> Matrix -> Matrix


-- | ReLu Activation.
module MachineLearning.NeuralNetwork.ReluActivation
relu :: Matrix -> Matrix
gradient :: Matrix -> Matrix -> Matrix


-- | Regularization.
module MachineLearning.NeuralNetwork.Regularization

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization

-- | Calcaulates regularization for forward propagation. It takes
--   regularization parameter and theta list.
forwardReg :: Regularization -> [(Matrix, Matrix)] -> R

-- | Calculates regularization for step of backward propagation. It takes
--   regularization parameter and theta.
backwardReg :: Regularization -> Matrix -> Matrix


-- | Regression Model type class.
module MachineLearning.Model
class Model a

-- | Hypothesis function, a.k.a. score function (for lassifition problem)
--   Takes X (m x n) and theta (n x 1), returns y (m x 1).
hypothesis :: Model a => a -> Matrix -> Vector -> Vector

-- | Cost function J(Theta), a.k.a. loss function. It takes regularizarion
--   parameter, matrix X (m x n), vector y (m x 1) and vector theta (n x
--   1).
cost :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> R

-- | Gradient function. It takes regularizarion parameter, X (m x n), y (m
--   x 1) and theta (n x 1). Returns vector of gradients (n x 1).
gradient :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> Vector


-- | Minibatch Gradient Descent
module MachineLearning.Optimization.MinibatchGradientDescent

-- | Minibatch Gradient Descent method implementation. See
--   <a>MachineLearning.Regression</a> for usage details.
minibatchGradientDescent :: Model a => Int -> Int -> R -> a -> R -> Int -> Regularization -> Matrix -> Vector -> Vector -> (Vector, Matrix)


module MachineLearning.Optimization.GradientDescent

-- | Gradient Descent method implementation. See
--   <a>MachineLearning.Regression</a> for usage details.
gradientDescent :: Model a => R -> a -> R -> Int -> Regularization -> Matrix -> Vector -> Vector -> (Vector, Matrix)


-- | Optimization module.
module MachineLearning.Optimization
data MinimizeMethod

-- | Gradient descent, takes alpha. Requires feature normalization.
GradientDescent :: R -> MinimizeMethod

-- | Minibacth Gradietn Descent, takes seed, batch size and alpha
MinibatchGradientDescent :: Int -> Int -> R -> MinimizeMethod

-- | Fletcher-Reeves conjugate gradient algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientFR :: R -> R -> MinimizeMethod

-- | Polak-Ribiere conjugate gradient algorithm. takes size of first trial
--   step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientPR :: R -> R -> MinimizeMethod

-- | Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
BFGS2 :: R -> R -> MinimizeMethod

-- | Returns solution vector (theta) and optimization path. Optimization
--   path's row format: [iter number, cost function value, theta values...]
minimize :: Model a => MinimizeMethod -> a -> R -> Int -> Regularization -> Matrix -> Vector -> Vector -> (Vector, Matrix)

-- | Gradient checking function. Approximates the derivates of the Model's
--   cost function and calculates derivatives using the Model's gradient
--   functions. Returns norn_2 between 2 derivatives. Takes model,
--   regularization, X, y, theta and epsilon (used to approximate
--   derivatives, 1e-4 is a good value).
checkGradient :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> R -> R


module MachineLearning.LogisticModel
data LogisticModel
Logistic :: LogisticModel

-- | Calculates sigmoid
sigmoid :: Floating a => a -> a

-- | Calculates derivatives of sigmoid
sigmoidGradient :: Floating a => a -> a
instance MachineLearning.Model.Model MachineLearning.LogisticModel.LogisticModel


-- | Sigmoid Activation.
module MachineLearning.NeuralNetwork.SigmoidActivation

-- | Calculates sigmoid
sigmoid :: Floating a => a -> a
gradient :: Matrix -> Matrix -> Matrix


-- | Logistic Loss.
module MachineLearning.NeuralNetwork.LogisticLoss
scores :: Matrix -> Matrix
gradient :: Matrix -> Matrix -> Matrix
loss :: Matrix -> Matrix -> R


module MachineLearning.LeastSquaresModel
data LeastSquaresModel
LeastSquares :: LeastSquaresModel
instance MachineLearning.Model.Model MachineLearning.LeastSquaresModel.LeastSquaresModel


module MachineLearning.Regression
class Model a

-- | Hypothesis function, a.k.a. score function (for lassifition problem)
--   Takes X (m x n) and theta (n x 1), returns y (m x 1).
hypothesis :: Model a => a -> Matrix -> Vector -> Vector

-- | Cost function J(Theta), a.k.a. loss function. It takes regularizarion
--   parameter, matrix X (m x n), vector y (m x 1) and vector theta (n x
--   1).
cost :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> R

-- | Gradient function. It takes regularizarion parameter, X (m x n), y (m
--   x 1) and theta (n x 1). Returns vector of gradients (n x 1).
gradient :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> Vector
data LeastSquaresModel
LeastSquares :: LeastSquaresModel
data MinimizeMethod

-- | Gradient descent, takes alpha. Requires feature normalization.
GradientDescent :: R -> MinimizeMethod

-- | Minibacth Gradietn Descent, takes seed, batch size and alpha
MinibatchGradientDescent :: Int -> Int -> R -> MinimizeMethod

-- | Fletcher-Reeves conjugate gradient algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientFR :: R -> R -> MinimizeMethod

-- | Polak-Ribiere conjugate gradient algorithm. takes size of first trial
--   step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientPR :: R -> R -> MinimizeMethod

-- | Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
BFGS2 :: R -> R -> MinimizeMethod

-- | Returns solution vector (theta) and optimization path. Optimization
--   path's row format: [iter number, cost function value, theta values...]
minimize :: Model a => MinimizeMethod -> a -> R -> Int -> Regularization -> Matrix -> Vector -> Vector -> (Vector, Matrix)

-- | Normal equation using inverse, does not require feature normalization
--   It takes X and y, returns theta.
normalEquation :: Matrix -> Vector -> Vector

-- | Normal equation using pseudo inverse, requires feature normalization
--   It takes X and y, returns theta.
normalEquation_p :: Matrix -> Vector -> Vector

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization


-- | Cluster Analysis a.k.a. Clustering - grouping data into coherent
--   subsets.
module MachineLearning.Clustering

-- | Cluster type (list of samples associtaed with the cluster).
type Cluster = Vector Vector

-- | Clusters data using K-Means Algorithm inside Random Monad. Runs
--   K-Means algorithm <tt>N</tt> times, returns the clustering with
--   smaller cost.
kmeans :: RandomGen g => Int -> Matrix -> Int -> Rand g (Vector Cluster)

-- | Run K-Means algorithm once inside Random Monad.
kmeansIterM :: RandomGen g => Vector Vector -> Int -> Int -> Rand g (Vector Cluster, [R])


-- | Learn function with progress bar for terminal.
module MachineLearning.TerminalProgress

-- | Learn the given function displaying progress bar in terminal. It takes
--   function, initial theta and number of iterations to call the function.
--   It returns theta and optimization path (see
--   <a>MachineLearning.Optimization</a> for details).
learnWithProgressBar :: (Vector -> (Vector, Matrix)) -> Vector -> Int -> IO (Vector, Matrix)

-- | Learn the given function displaying progress bar in terminal. It takes
--   function, list of outputs and list of initial thetas and number of
--   iterations to call the function. It returns list of thetas and list of
--   optimization paths (see <a>MachineLearning.Optimization</a> for
--   details).
learnOneVsAllWithProgressBar :: (Vector -> Vector -> (Vector, Matrix)) -> Vector -> [Vector] -> Int -> IO ([Vector], [Matrix])


-- | One-vs-All Classification.
module MachineLearning.Classification.OneVsAll
data MinimizeMethod

-- | Gradient descent, takes alpha. Requires feature normalization.
GradientDescent :: R -> MinimizeMethod

-- | Minibacth Gradietn Descent, takes seed, batch size and alpha
MinibatchGradientDescent :: Int -> Int -> R -> MinimizeMethod

-- | Fletcher-Reeves conjugate gradient algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientFR :: R -> R -> MinimizeMethod

-- | Polak-Ribiere conjugate gradient algorithm. takes size of first trial
--   step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientPR :: R -> R -> MinimizeMethod

-- | Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
BFGS2 :: R -> R -> MinimizeMethod

-- | One-vs-All Classification prediction function. Takes a matrix of
--   features X and a list of vectors theta, returns predicted class number
--   assuming that class numbers start at 0.
predict :: Matrix -> [Vector] -> Vector

-- | Learns One-vs-All Classification
learn :: MinimizeMethod -> R -> Int -> Regularization -> Int -> Matrix -> Vector -> [Vector] -> ([Vector], [Matrix])

-- | Calculates accuracy of Classification predictions. Takes vector
--   expected y and vector predicted y. Returns number from 0 to 1, the
--   closer to 1 the better accuracy. Suitable for both Classification
--   Types: Binary and Multiclass.
calcAccuracy :: Vector -> Vector -> R

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization


-- | Binary Classification.
module MachineLearning.Classification.Binary
data MinimizeMethod

-- | Gradient descent, takes alpha. Requires feature normalization.
GradientDescent :: R -> MinimizeMethod

-- | Minibacth Gradietn Descent, takes seed, batch size and alpha
MinibatchGradientDescent :: Int -> Int -> R -> MinimizeMethod

-- | Fletcher-Reeves conjugate gradient algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientFR :: R -> R -> MinimizeMethod

-- | Polak-Ribiere conjugate gradient algorithm. takes size of first trial
--   step (0.1 is fine) and tol (0.1 is fine).
ConjugateGradientPR :: R -> R -> MinimizeMethod

-- | Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm, takes size of first
--   trial step (0.1 is fine) and tol (0.1 is fine).
BFGS2 :: R -> R -> MinimizeMethod

-- | Binary Classification prediction function. Takes a matrix of features
--   X and a vector theta. Returns predicted class.
predict :: Matrix -> Vector -> Vector

-- | Learns Binary Classification.
learn :: MinimizeMethod -> R -> Int -> Regularization -> Matrix -> Vector -> Vector -> (Vector, Matrix)

-- | Calculates accuracy of Classification predictions. Takes vector
--   expected y and vector predicted y. Returns number from 0 to 1, the
--   closer to 1 the better accuracy. Suitable for both Classification
--   Types: Binary and Multiclass.
calcAccuracy :: Vector -> Vector -> R

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization


module MachineLearning

-- | Add biad dimension to the future matrix
addBiasDimension :: Matrix -> Matrix

-- | Remove biad dimension
removeBiasDimension :: Matrix -> Matrix

-- | Caclulates mean and stddev values of every feature. Takes feature
--   matrix X, returns pair of vectors of means and stddevs.
meanStddev :: Matrix R -> (Matrix R, Matrix R)
featureNormalization :: Fractional a => (a, a) -> a -> a

-- | Maps the features into all polynomial terms of X up to the degree-th
--   power
mapFeatures :: Int -> Matrix -> Matrix

-- | Splits data matrix to features matrix X and vector of outputs y
splitToXY :: Element t => Matrix t -> (Matrix t, Vector t)


-- | Principal Component Analysis (PCA) - dimensionality reduction
--   algorithm. It is mostly used to speed up supervising learning
--   (Regression, Classification, etc) and visualization of data.
module MachineLearning.PCA

-- | Gets dimensionality reduction function, retained variance (0..1) and
--   reduced X for given matrix X and number of dimensions to retain.
getDimReducer :: Matrix -> Int -> (Matrix -> Matrix, R, Matrix)

-- | Gets dimensionality reduction function, retained number of dimensions
--   and reduced X for given matrix X and variance to retain (0..1].
getDimReducer_rv :: Matrix -> R -> (Matrix -> Matrix, Int, Matrix)


-- | MultiClass Classification.
module MachineLearning.Classification.MultiClass

-- | Classifier type class represents Multi-class classification models.
class Classifier a

-- | Score function
cscore :: Classifier a => a -> Matrix -> Matrix -> Matrix

-- | Hypothesis function Takes X (m x n) and theta (n x k), returns y (m x
--   k).
chypothesis :: Classifier a => a -> Matrix -> Matrix -> Vector

-- | Cost function J(Theta), a.k.a. loss function. It takes regularizarion
--   parameter lambda, matrix X (m x n), vector y (m x 1) and vector theta
--   (n x 1).
ccost :: Classifier a => a -> Regularization -> Matrix -> Vector -> Matrix -> R

-- | Gradient function. It takes regularizarion parameter lambda, X (m x
--   n), y (m x 1) and theta (n x 1). Returns vector of gradients (n x 1).
cgradient :: Classifier a => a -> Regularization -> Matrix -> Vector -> Matrix -> Matrix

-- | Returns Number of Classes
cnumClasses :: Classifier a => a -> Int

-- | MultiClassModel is Model wrapper class around Classifier
data MultiClassModel m
MultiClass :: m -> MultiClassModel m

-- | Process outputs for MultiClass Classification. Takes Classifier and
--   output vector y. Returns matrix of binary outputs. It is supposed that
--   labels are integerets start at 0.
processOutput :: (Classifier c) => c -> Vector -> Matrix

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization

-- | Calculates regularization for Classifier.ccost. It takes
--   regularization parameter and theta.
ccostReg :: Regularization -> Matrix -> R

-- | Calculates regularization for Classifier.cgradient. It takes
--   regularization parameter and theta.
cgradientReg :: Regularization -> Matrix -> Matrix
instance MachineLearning.Classification.MultiClass.Classifier a => MachineLearning.Model.Model (MachineLearning.Classification.MultiClass.MultiClassModel a)


-- | Various helpful utilities.
module MachineLearning.Utils
reduceByRowsV :: (Vector -> R) -> Matrix -> Vector
reduceByColumnsV :: (Vector -> R) -> Matrix -> Vector
reduceByRows :: (Vector -> R) -> Matrix -> Matrix
reduceByColumns :: (Vector -> R) -> Matrix -> Matrix
sumByRows :: Matrix -> Matrix
sumByColumns :: Matrix -> Matrix

-- | Converts list of tuples into list.
listOfTuplesToList :: [(a, a)] -> [a]


-- | Softmax Classifier (Multiclass Logistic Regression).
module MachineLearning.SoftmaxClassifier

-- | Softmax Classifier, takes number of classes.
data SoftmaxClassifier
Softmax :: Int -> SoftmaxClassifier
instance MachineLearning.Classification.MultiClass.Classifier MachineLearning.SoftmaxClassifier.SoftmaxClassifier


-- | Softmax Loss.
module MachineLearning.NeuralNetwork.SoftmaxLoss
scores :: Matrix R -> Matrix R
gradient :: Matrix R -> Matrix R -> Matrix R
loss :: Matrix -> Matrix -> R


-- | Multi SVM Loss.
module MachineLearning.NeuralNetwork.MultiSvmLoss
scores :: () => a -> a
gradient :: Matrix -> Matrix -> Matrix
loss :: Matrix -> Matrix -> R


-- | Neural Network's Layer
module MachineLearning.NeuralNetwork.Layer
data Layer
Layer :: Int -> (Matrix -> Matrix -> Matrix -> Matrix) -> (Matrix -> Cache -> (Matrix, Matrix, Matrix)) -> (Matrix -> Matrix) -> (Matrix -> Matrix -> Matrix) -> (forall g. RandomGen g => (Int, Int) -> Rand g (Matrix, Matrix)) -> Layer
[lUnits] :: Layer -> Int
[lForward] :: Layer -> Matrix -> Matrix -> Matrix -> Matrix
[lBackward] :: Layer -> Matrix -> Cache -> (Matrix, Matrix, Matrix)
[lActivation] :: Layer -> Matrix -> Matrix
[lActivationGradient] :: Layer -> Matrix -> Matrix -> Matrix
[lInitializeThetaM] :: Layer -> forall g. RandomGen g => (Int, Int) -> Rand g (Matrix, Matrix)
data Cache
Cache :: Matrix -> Matrix -> Matrix -> Cache
[cacheZ] :: Cache -> Matrix
[cacheX] :: Cache -> Matrix
[cacheW] :: Cache -> Matrix
affineForward :: Matrix -> Matrix -> Matrix -> Matrix
affineBackward :: Matrix R -> Cache -> (Matrix R, Matrix, Matrix R)


-- | Neural Network's Topology
module MachineLearning.NeuralNetwork.Topology

-- | Neural network topology has at least 2 elements: numver of input and
--   number of outputs. And sizes of hidden layers between 2 elements.
data Topology

-- | Loss function's type. Takes x, weights and y.
type LossFunc = Matrix -> Matrix -> R

-- | Makes Neural Network's Topology. Takes number of inputs, list of
--   hidden layers, output layer and loss function.
makeTopology :: Int -> [Layer] -> Layer -> LossFunc -> Topology

-- | Calculates loss for the given topology. Takes topology,
--   regularization, x, weights, y.
loss :: Topology -> Regularization -> Matrix -> [(Matrix, Matrix)] -> Matrix -> R

-- | Implementation of forward propagation algorithm.
propagateForward :: Topology -> Matrix -> [(Matrix, Matrix)] -> (Matrix, [Cache])

-- | Implementation of backward propagation algorithm.
propagateBackward :: Topology -> Regularization -> Matrix -> [Cache] -> Matrix -> [(Matrix, Matrix)]

-- | Returns number of outputs of the given topology.
numberOutputs :: Topology -> Int

-- | Create and initialize weights vector with random values for given
--   neural network topology. Takes a seed to initialize generator of
--   random numbers as a first parameter.
initializeTheta :: Int -> Topology -> Vector

-- | Create and initialize weights vector with random values for given
--   neural network topology inside IO Monad.
initializeThetaIO :: Topology -> IO Vector

-- | Create and initialize weights vector with random values for given
--   neural network topology inside RandomMonad.
initializeThetaM :: RandomGen g => Topology -> Rand g Vector

-- | Flatten list of matrices into vector.
flatten :: [(Matrix, Matrix)] -> Vector

-- | Unflatten vector into list of matrices for given neural network
--   topology.
unflatten :: Topology -> Vector -> [(Matrix, Matrix)]


-- | Topology Maker.
module MachineLearning.NeuralNetwork.TopologyMaker
data Activation
ASigmoid :: Activation
ARelu :: Activation
ATanh :: Activation
data Loss
LLogistic :: Loss
LSoftmax :: Loss
LMultiSvm :: Loss

-- | Creates toplogy. Takes number of inputs, number of outputs and list of
--   hidden layers.
makeTopology :: Activation -> Loss -> Int -> Int -> [Int] -> Topology


-- | Simple Neural Networks.
module MachineLearning.NeuralNetwork
class Model a

-- | Hypothesis function, a.k.a. score function (for lassifition problem)
--   Takes X (m x n) and theta (n x 1), returns y (m x 1).
hypothesis :: Model a => a -> Matrix -> Vector -> Vector

-- | Cost function J(Theta), a.k.a. loss function. It takes regularizarion
--   parameter, matrix X (m x n), vector y (m x 1) and vector theta (n x
--   1).
cost :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> R

-- | Gradient function. It takes regularizarion parameter, X (m x n), y (m
--   x 1) and theta (n x 1). Returns vector of gradients (n x 1).
gradient :: Model a => a -> Regularization -> Matrix -> Vector -> Vector -> Vector

-- | Neural Network Model. Takes neural network topology as a constructor
--   argument.
newtype NeuralNetworkModel
NeuralNetwork :: Topology -> NeuralNetworkModel

-- | Calculates accuracy of Classification predictions. Takes vector
--   expected y and vector predicted y. Returns number from 0 to 1, the
--   closer to 1 the better accuracy. Suitable for both Classification
--   Types: Binary and Multiclass.
calcAccuracy :: Vector -> Vector -> R

-- | Neural network topology has at least 2 elements: numver of input and
--   number of outputs. And sizes of hidden layers between 2 elements.
data Topology

-- | Create and initialize weights vector with random values for given
--   neural network topology. Takes a seed to initialize generator of
--   random numbers as a first parameter.
initializeTheta :: Int -> Topology -> Vector

-- | Create and initialize weights vector with random values for given
--   neural network topology inside IO Monad.
initializeThetaIO :: Topology -> IO Vector

-- | Create and initialize weights vector with random values for given
--   neural network topology inside RandomMonad.
initializeThetaM :: RandomGen g => Topology -> Rand g Vector

-- | Regularization
data Regularization

-- | No regularization
RegNone :: Regularization

-- | L2 Regularization
L2 :: R -> Regularization
instance MachineLearning.Model.Model MachineLearning.NeuralNetwork.NeuralNetworkModel


-- | Multicalss Support Vector Machines Classifier.
module MachineLearning.MultiSvmClassifier

-- | Multiclass SVM Classifier, takes delta and number of futures. Delta =
--   1.0 is good for all cases.
data MultiSvmClassifier
MultiSvm :: R -> Int -> MultiSvmClassifier
instance MachineLearning.Classification.MultiClass.Classifier MachineLearning.MultiSvmClassifier.MultiSvmClassifier