Uploaded v1.1

mikeizbicki · Jul 27, 2013 · 89ce26a · 89ce26a
1 parent 2a2501b
commit 89ce26a
Show file tree

Hide file tree

Showing 16 changed files with 193 additions and 1,677 deletions.
diff --git a/HLearn-algebra/HLearn-algebra.cabal b/HLearn-algebra/HLearn-algebra.cabal
@@ -15,7 +15,7 @@ bug-reports:         http://github.com/mikeizbicki/HLearn/issues
 Library
     Build-Depends:      
         base                >= 3 && < 5,
-        ConstraintKinds     >= 0.1.0,
+        ConstraintKinds     >= 1.1.0,
         vector-heterogenous >= 0.1.1,
         template-haskell,
 
@@ -57,9 +57,9 @@ Library
         BangPatterns
         TypeOperators
         GeneralizedNewtypeDeriving
-        DataKinds
+        --DataKinds
         TypeFamilies
-        PolyKinds
+        --PolyKinds
         StandaloneDeriving
         GADTs
         KindSignatures

diff --git a/HLearn-algebra/src/HLearn/Algebra/Models/HomTrainer.hs b/HLearn-algebra/src/HLearn/Algebra/Models/HomTrainer.hs
@@ -10,6 +10,12 @@ module HLearn.Algebra.Models.HomTrainer
     HomTrainer (..)
     , WeightedHomTrainer (..)
     , NumDP(..)
+
+    -- * useful functions
+    , sub1dp
+    , subBatch
+    , sub1dpW
+    , subBatchW
 
     )
     where
@@ -88,6 +94,8 @@ class (Monoid model) => HomTrainer model where
 -------------------------------------------------------------------------------
 -- WeightedHomTrainer
 
+type WeightedDatapoint model = (Ring model, Datapoint model)
+
 class (Module model, HomTrainer model) => 
     WeightedHomTrainer model 
         where
@@ -99,12 +107,32 @@ class (Module model, HomTrainer model) =>
         container (Ring model,Datapoint model) -> model
     trainW = batch train1dpW
 
-    add1dpW :: model -> (Ring model,Datapoint model) -> model
+    add1dpW :: model -> WeightedDatapoint model -> model
     add1dpW = online $ unbatch $ offline addBatchW
 
     addBatchW :: (Foldable container, Functor container) => 
-        model -> container (Ring model,Datapoint model) -> model
+        model -> container (WeightedDatapoint model) -> model
     addBatchW = online trainW
 
 instance (Module model, HomTrainer model) => WeightedHomTrainer model
-
+
+-------------------------------------------------------------------------------
+-- helper functions
+
+-- | subtracts a single data point from the model
+sub1dp :: (Group model, HomTrainer model) => model -> Datapoint model -> model
+sub1dp m dp = m <> inverse (train1dp dp)
+
+-- | subtracts a multiple data point from the model
+subBatch :: (Group model, HomTrainer model, Foldable container, Functor container) => 
+    model -> container (Datapoint model) -> model
+subBatch m dpL = m <> inverse (train dpL)
+
+-- | subtracts a single weighted data point from the model
+sub1dpW :: (Group model, WeightedHomTrainer model) => model -> WeightedDatapoint model -> model
+sub1dpW m dp = m <> inverse (train1dpW dp)
+
+-- | subtracts multiple weighted data points from the model
+subBatchW :: (Group model, WeightedHomTrainer model, Foldable container, Functor container) => 
+    model -> container (WeightedDatapoint model) -> model
+subBatchW m dpL = m <> inverse (trainW dpL)
diff --git a/HLearn-algebra/src/HLearn/Algebra/Structures/Free/Bagging.hs b/HLearn-algebra/src/HLearn/Algebra/Structures/Free/Bagging.hs
@@ -1,3 +1,5 @@
+{-# LANGUAGE DataKinds #-}
+
 module HLearn.Algebra.Structures.Free.Bagging
     ( Bagging
     , Bagging'
@@ -57,4 +59,4 @@ instance
         where
             n = fromIntegral $ fromSing (sing :: Sing n)
             seed = fromIntegral $ fromSing (sing :: Sing seed)
-            (offset,g) = random $ mkStdGen $ hash dp + seed
+            (offset,g) = random $ mkStdGen $ hash dp + seed
diff --git a/HLearn-algebra/src/HLearn/Algebra/Structures/Free/FreeHomTrainer.hs b/HLearn-algebra/src/HLearn/Algebra/Structures/Free/FreeHomTrainer.hs
@@ -1,4 +1,5 @@
 {-# LANGUAGE EmptyDataDecls #-}
+{-# LANGUAGE PolyKinds #-}
 
 module HLearn.Algebra.Structures.Free.FreeHomTrainer
     ( FreeHomTrainer (..)
@@ -71,4 +72,4 @@ instance
 
     train1dp dp = FreeHomTrainer'
         { modelL = FreeModule $ Map.singleton (lame_train dp) 1
-        }
+        }
diff --git a/HLearn-algebra/src/HLearn/Algebra/Types/Indexing.hs b/HLearn-algebra/src/HLearn/Algebra/Types/Indexing.hs
@@ -1,3 +1,4 @@
+{-# LANGUAGE DataKinds #-}
 module HLearn.Algebra.Types.Indexing
     ( 
     Index(..)

diff --git a/HLearn-algebra/src/HLearn/Algebra/Types/Nat1.hs b/HLearn-algebra/src/HLearn/Algebra/Types/Nat1.hs
diff --git a/HLearn-algebra/src/HLearn/Algebra/Types/Nat1Generator.hs b/HLearn-algebra/src/HLearn/Algebra/Types/Nat1Generator.hs
@@ -4,7 +4,8 @@ data Nat1 = Zero | Succ Nat1
 
 decl :: String
 decl = concat $ map (\x -> x++"\n") 
-    [ "-- | This module was automatically generated by Nat1Generator.hs"
+    [ "{-# LANGUAGE DataKinds #-}"
+    , "-- | This module was automatically generated by Nat1Generator.hs"
     , "module HLearn.Algebra.Types.Nat1 where"
     , "import GHC.TypeLits"
     , "data Nat1Box (n::Nat1) = Nat1Box"
@@ -26,7 +27,7 @@ string_FromNat1 :: Int -> String
 string_FromNat1 i = "type instance FromNat1 ("++show (int2Nat1 i)++") = "++show i
 
 numNat :: Int
-numNat = 1000
+numNat = 200
 
 main = do
     putStrLn decl

diff --git a/HLearn-distributions/HLearn-distributions.cabal b/HLearn-distributions/HLearn-distributions.cabal
@@ -85,9 +85,9 @@ Library
         BangPatterns
         TypeOperators
         GeneralizedNewtypeDeriving
-        DataKinds
+        --DataKinds
         TypeFamilies
-        PolyKinds
+        --PolyKinds
         StandaloneDeriving
         GADTs
         KindSignatures

diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/CatContainer.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/CatContainer.hs
@@ -1,3 +1,4 @@
+{-# LANGUAGE DataKinds #-}
 -- | The categorical distribution is used for discrete data.  It is also sometimes called the discrete distribution or the multinomial distribution.  For more, see the wikipedia entry: <https://en.wikipedia.org/wiki/CatContainer_distribution>
 module HLearn.Models.Distributions.Multivariate.Internal.CatContainer
 {-    ( 
@@ -217,4 +218,4 @@ ds= [ "test":::'g':::"foo":::HNil
     , "toot":::'f':::"foo":::HNil
     ]
 
-test = train ds :: CatContainer String (CatContainer Char (CatContainer String (Unital Double) Double) Double) Double
+test = train ds :: CatContainer String (CatContainer Char (CatContainer String (Unital Double) Double) Double) Double
diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/Container.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/Container.hs
@@ -1,3 +1,5 @@
+{-# LANGUAGE DataKinds #-}
+{-# LANGUAGE PolyKinds #-}
 module HLearn.Models.Distributions.Multivariate.Internal.Container
     ( Container
     , MultiContainer

diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/Ignore.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/Ignore.hs
@@ -1,3 +1,4 @@
+{-# LANGUAGE DataKinds #-}
 -- | Used for ignoring data
 module HLearn.Models.Distributions.Multivariate.Internal.Ignore
     ( Ignore 
@@ -85,4 +86,4 @@ instance
         Ignore' label (MarginalizeOut' (Nat1Box n) basedist) prob
     marginalizeOut' _ dist = Ignore' $ marginalizeOut' (undefined :: Nat1Box n) $ basedist dist
 
-    condition' _ dist dp = Ignore' $ condition' (undefined :: Nat1Box n) (basedist dist) dp
+    condition' _ dist dp = Ignore' $ condition' (undefined :: Nat1Box n) (basedist dist) dp
diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/TypeLens.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/TypeLens.hs
@@ -1,3 +1,5 @@
+{-# LANGUAGE DataKinds #-}
+
 -- | This module provides convenient TemplateHaskell functions for making type lens suitable for use with multivariate distributions.
 -- 
 -- Given a data type that looks like:

diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/Unital.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Multivariate/Internal/Unital.hs
@@ -1,3 +1,4 @@
+{-# LANGUAGE DataKinds #-}
 module HLearn.Models.Distributions.Multivariate.Internal.Unital
     where
 

diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Univariate/Internal/MissingData.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Univariate/Internal/MissingData.hs
@@ -1,3 +1,4 @@
+{-# LANGUAGE DataKinds #-}
 -- | Adapts any distribution into one that can handle missing data
 module HLearn.Models.Distributions.Univariate.Internal.MissingData
     ( MissingData

diff --git a/HLearn-distributions/src/HLearn/Models/Distributions/Univariate/KernelDensityEstimator.hs b/HLearn-distributions/src/HLearn/Models/Distributions/Univariate/KernelDensityEstimator.hs
@@ -1,3 +1,4 @@
+{-# LANGUAGE DataKinds #-}
 -- | Kernel Density Estimation (KDE) is a generic and powerful method for estimating a probability distribution.  See wikipedia for more information: <http://en.wikipedia.org/wiki/Kernel_density_estimation>
 module HLearn.Models.Distributions.Univariate.KernelDensityEstimator
     where