data wrapper is added

jumutc · Jul 16, 2015 · 768c1a1 · 768c1a1
1 parent 0942edb
commit 768c1a1
Show file tree

Hide file tree

Showing 20 changed files with 264 additions and 42 deletions.
diff --git a/REQUIRE b/REQUIRE
@@ -2,4 +2,6 @@ MAT
 MLBase
 Compat
 Distances
-Distributions
+Distributions
+DataArrays
+DataFrames
diff --git a/data/iris.data.csv b/data/iris.data.csv
@@ -0,0 +1,150 @@
+5.1,3.5,1.4,0.2,1
+4.9,3,1.4,0.2,1
+4.7,3.2,1.3,0.2,1
+4.6,3.1,1.5,0.2,1
+5,3.6,1.4,0.2,1
+5.4,3.9,1.7,0.4,1
+4.6,3.4,1.4,0.3,1
+5,3.4,1.5,0.2,1
+4.4,2.9,1.4,0.2,1
+4.9,3.1,1.5,0.1,1
+5.4,3.7,1.5,0.2,1
+4.8,3.4,1.6,0.2,1
+4.8,3,1.4,0.1,1
+4.3,3,1.1,0.1,1
+5.8,4,1.2,0.2,1
+5.7,4.4,1.5,0.4,1
+5.4,3.9,1.3,0.4,1
+5.1,3.5,1.4,0.3,1
+5.7,3.8,1.7,0.3,1
+5.1,3.8,1.5,0.3,1
+5.4,3.4,1.7,0.2,1
+5.1,3.7,1.5,0.4,1
+4.6,3.6,1,0.2,1
+5.1,3.3,1.7,0.5,1
+4.8,3.4,1.9,0.2,1
+5,3,1.6,0.2,1
+5,3.4,1.6,0.4,1
+5.2,3.5,1.5,0.2,1
+5.2,3.4,1.4,0.2,1
+4.7,3.2,1.6,0.2,1
+4.8,3.1,1.6,0.2,1
+5.4,3.4,1.5,0.4,1
+5.2,4.1,1.5,0.1,1
+5.5,4.2,1.4,0.2,1
+4.9,3.1,1.5,0.1,1
+5,3.2,1.2,0.2,1
+5.5,3.5,1.3,0.2,1
+4.9,3.1,1.5,0.1,1
+4.4,3,1.3,0.2,1
+5.1,3.4,1.5,0.2,1
+5,3.5,1.3,0.3,1
+4.5,2.3,1.3,0.3,1
+4.4,3.2,1.3,0.2,1
+5,3.5,1.6,0.6,1
+5.1,3.8,1.9,0.4,1
+4.8,3,1.4,0.3,1
+5.1,3.8,1.6,0.2,1
+4.6,3.2,1.4,0.2,1
+5.3,3.7,1.5,0.2,1
+5,3.3,1.4,0.2,1
+7,3.2,4.7,1.4,2
+6.4,3.2,4.5,1.5,2
+6.9,3.1,4.9,1.5,2
+5.5,2.3,4,1.3,2
+6.5,2.8,4.6,1.5,2
+5.7,2.8,4.5,1.3,2
+6.3,3.3,4.7,1.6,2
+4.9,2.4,3.3,1,2
+6.6,2.9,4.6,1.3,2
+5.2,2.7,3.9,1.4,2
+5,2,3.5,1,2
+5.9,3,4.2,1.5,2
+6,2.2,4,1,2
+6.1,2.9,4.7,1.4,2
+5.6,2.9,3.6,1.3,2
+6.7,3.1,4.4,1.4,2
+5.6,3,4.5,1.5,2
+5.8,2.7,4.1,1,2
+6.2,2.2,4.5,1.5,2
+5.6,2.5,3.9,1.1,2
+5.9,3.2,4.8,1.8,2
+6.1,2.8,4,1.3,2
+6.3,2.5,4.9,1.5,2
+6.1,2.8,4.7,1.2,2
+6.4,2.9,4.3,1.3,2
+6.6,3,4.4,1.4,2
+6.8,2.8,4.8,1.4,2
+6.7,3,5,1.7,2
+6,2.9,4.5,1.5,2
+5.7,2.6,3.5,1,2
+5.5,2.4,3.8,1.1,2
+5.5,2.4,3.7,1,2
+5.8,2.7,3.9,1.2,2
+6,2.7,5.1,1.6,2
+5.4,3,4.5,1.5,2
+6,3.4,4.5,1.6,2
+6.7,3.1,4.7,1.5,2
+6.3,2.3,4.4,1.3,2
+5.6,3,4.1,1.3,2
+5.5,2.5,4,1.3,2
+5.5,2.6,4.4,1.2,2
+6.1,3,4.6,1.4,2
+5.8,2.6,4,1.2,2
+5,2.3,3.3,1,2
+5.6,2.7,4.2,1.3,2
+5.7,3,4.2,1.2,2
+5.7,2.9,4.2,1.3,2
+6.2,2.9,4.3,1.3,2
+5.1,2.5,3,1.1,2
+5.7,2.8,4.1,1.3,2
+6.3,3.3,6,2.5,3
+5.8,2.7,5.1,1.9,3
+7.1,3,5.9,2.1,3
+6.3,2.9,5.6,1.8,3
+6.5,3,5.8,2.2,3
+7.6,3,6.6,2.1,3
+4.9,2.5,4.5,1.7,3
+7.3,2.9,6.3,1.8,3
+6.7,2.5,5.8,1.8,3
+7.2,3.6,6.1,2.5,3
+6.5,3.2,5.1,2,3
+6.4,2.7,5.3,1.9,3
+6.8,3,5.5,2.1,3
+5.7,2.5,5,2,3
+5.8,2.8,5.1,2.4,3
+6.4,3.2,5.3,2.3,3
+6.5,3,5.5,1.8,3
+7.7,3.8,6.7,2.2,3
+7.7,2.6,6.9,2.3,3
+6,2.2,5,1.5,3
+6.9,3.2,5.7,2.3,3
+5.6,2.8,4.9,2,3
+7.7,2.8,6.7,2,3
+6.3,2.7,4.9,1.8,3
+6.7,3.3,5.7,2.1,3
+7.2,3.2,6,1.8,3
+6.2,2.8,4.8,1.8,3
+6.1,3,4.9,1.8,3
+6.4,2.8,5.6,2.1,3
+7.2,3,5.8,1.6,3
+7.4,2.8,6.1,1.9,3
+7.9,3.8,6.4,2,3
+6.4,2.8,5.6,2.2,3
+6.3,2.8,5.1,1.5,3
+6.1,2.6,5.6,1.4,3
+7.7,3,6.1,2.3,3
+6.3,3.4,5.6,2.4,3
+6.4,3.1,5.5,1.8,3
+6,3,4.8,1.8,3
+6.9,3.1,5.4,2.1,3
+6.7,3.1,5.6,2.4,3
+6.9,3.1,5.1,2.3,3
+5.8,2.7,5.1,1.9,3
+6.8,3.2,5.9,2.3,3
+6.7,3.3,5.7,2.5,3
+6.7,3,5.2,2.3,3
+6.3,2.5,5,1.9,3
+6.5,3,5.2,2,3
+6.2,3.4,5.4,2.3,3
+5.9,3,5.1,1.8,3
diff --git a/src/SALSA.jl b/src/SALSA.jl
@@ -1,8 +1,9 @@
 module SALSA
 
-export salsa, mapstd, make_sparse, entropysubset, AFEm, gen_cross_validate
+export salsa, mapstd, make_sparse, entropy_subset, AFEm, gen_cross_validate
 
 using MLBase, Distributions, Compat, Distances
+import Base: size, getindex, issparse, sub
 
 
 # Calculate the misclassification rate
@@ -29,7 +30,8 @@ end
 # needed support files
 include(joinpath("kernels", "kernels.jl"))
 include(joinpath("support", "constants.jl"))
-include(joinpath("support", "entropysubset.jl"))
+include(joinpath("support", "entropy_subset.jl"))
+include(joinpath("support", "data_wrapper.jl"))
 include(joinpath("support", "sparse.jl"))
 include(joinpath("support", "mapstd.jl"))
 include(joinpath("support", "AFEm.jl"))
@@ -79,7 +81,7 @@ function gen_cross_validate(evalfun::Function, X, Y, model::SALSAModel)
 	indices = get(model.cv_gen, Kfold(length(Y),nfolds()))
     @parallel (+) for train_idx in collect(indices)
 		val_idx = setdiff(1:length(Y), train_idx)
-        evalfun(X[train_idx,:], Y[train_idx], X[val_idx,:], Y[val_idx])/nfolds()
+        evalfun(sub(X,train_idx,:), Y[train_idx], sub(X,val_idx,:), Y[val_idx])/nfolds()
     end
 end
 
@@ -123,9 +125,9 @@ function salsa(X, Y, model::SALSAModel, Xtest)
 		model.output = OutputModel{model.mode}()
 	end
 
-    if model.normalized && isempty(Xtest) 
+    if model.normalized && isempty(Xtest) && typeof(X) <: Array
 	    (X, model.output.X_mean, model.output.X_std) = mapstd(X)
-	elseif model.normalized
+	elseif model.normalized && typeof(X) <: Array
 	    (X, model.output.X_mean, model.output.X_std) = mapstd(X)
 	    Xtest = mapstd(Xtest,model.output.X_mean,model.output.X_std)
 	end

diff --git a/src/algorithms/adaptive_l1rda_alg.jl b/src/algorithms/adaptive_l1rda_alg.jl
@@ -16,13 +16,14 @@ function adaptive_l1rda_alg(dfunc::Function, X, Y, λ::Float64, γ::Float64, ρ:
         g = zeros(d,1)
         h = zeros(d,1)
         w = rand(d,1)/100
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else 
         g = spzeros(d,1)
         h = spzeros(d,1)
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))/100
-        A = [X'; sparse(ones(1,N))]
+        X = [X'; sparse(ones(1,N))]
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -48,7 +49,7 @@ function adaptive_l1rda_alg(dfunc::Function, X, Y, λ::Float64, γ::Float64, ρ:
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
+        At = sub_arr(idx)
 
         # calculate dual average: (cumulative) gradient
         g_new = dfunc(At,yt,w)

diff --git a/src/algorithms/dropout_alg.jl b/src/algorithms/dropout_alg.jl
@@ -13,12 +13,13 @@ function dropout_alg(dfunc::Function, X, Y, λ::Float64, k::Int, max_iter::Int,
     if ~check
         w = rand(d)
         rw = ones(d)/d 
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else 
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))
-        A = [X'; sparse(ones(1,N))]
+        X = [X'; sparse(ones(1,N))]
         f_sample = (p) -> rand(Bernoulli(p^2/(1+p^2))) 
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -44,7 +45,7 @@ function dropout_alg(dfunc::Function, X, Y, λ::Float64, k::Int, max_iter::Int,
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
+        At = sub_arr(idx)
 
         # define samples
         if ~check

diff --git a/src/algorithms/l1rda_alg.jl b/src/algorithms/l1rda_alg.jl
@@ -15,12 +15,13 @@ function l1rda_alg(dfunc::Function, X, Y, λ::Float64, γ::Float64, ρ::Float64,
     if check
         g = zeros(d,1)
         w = rand(d,1)/100
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else 
         g = spzeros(d,1)
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))/100
-        A = [X'; sparse(ones(1,N))]
+        X = [X'; sparse(ones(1,N))]
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -46,8 +47,8 @@ function l1rda_alg(dfunc::Function, X, Y, λ::Float64, γ::Float64, ρ::Float64,
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
-
+        At = sub_arr(idx)
+        
         # calculate dual average: gradient
         g = ((t-1)/t).*g + (1/(t)).*dfunc(At,yt,w)
         λ_rda = λ+(ρ*γ)/sqrt(t)

diff --git a/src/algorithms/pegasos_alg.jl b/src/algorithms/pegasos_alg.jl
@@ -13,12 +13,13 @@ function pegasos_alg(dfunc::Function, X, Y, λ::Float64, k::Int, max_iter::Int,
     if ~check
         w = rand(d)
         w = w./(sqrt(λ)*vecnorm(w))
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else 
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))
         w = w./(sqrt(λ)*vecnorm(w))
-        A = [X'; sparse(ones(1,N))]
+        X = [X'; sparse(ones(1,N))]
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -44,12 +45,12 @@ function pegasos_alg(dfunc::Function, X, Y, λ::Float64, k::Int, max_iter::Int,
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
+        At = sub_arr(idx)
 
         # do a gradient descent step
         η_t = 1/(λ*t)
         w = (1 - η_t*λ).*w
-        w = w - (η_t/k).*dfunc(At,yt,w_prev)#*class_weight[yt]
+        w = w - (η_t/k).*dfunc(At,yt,w_prev)
         # project back to the set B: w \in convex set B
         w = min(1,1/(sqrt(λ)*vecnorm(w))).*w
 

diff --git a/src/algorithms/reweighted_l1rda_alg.jl b/src/algorithms/reweighted_l1rda_alg.jl
@@ -17,12 +17,13 @@ function reweighted_l1rda_alg(dfunc::Function, X, Y, λ::Float64, γ::Float64,
         g = zeros(d)
         rw = ones(d)
         w = rand(d,1)/100
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else 
         g = spzeros(d,1)
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))/100
-        A = [X'; sparse(ones(1,N))]
+        X = [X'; sparse(ones(1,N))]
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -48,7 +49,7 @@ function reweighted_l1rda_alg(dfunc::Function, X, Y, λ::Float64, γ::Float64,
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
+        At = sub_arr(idx)
 
         # calculate dual average: gradient
         g = ((t-1)/t).*g + (1/(t)).*dfunc(At,yt,w)

diff --git a/src/algorithms/reweighted_l2rda_alg.jl b/src/algorithms/reweighted_l2rda_alg.jl
@@ -16,13 +16,14 @@ function reweighted_l2rda_alg(dfunc::Function, X, Y, λ::Float64, ɛ::Float64, v
     if check
         g = zeros(d,1)
         w = rand(d,1)/100
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else      
         g = spzeros(d,1)
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))/100
-        A = [X'; sparse(ones(1,N))]
         fg = i -> -1./(λ .+ rw[i])
+        X = [X'; sparse(ones(1,N))]
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -48,7 +49,7 @@ function reweighted_l2rda_alg(dfunc::Function, X, Y, λ::Float64, ɛ::Float64, v
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
+        At = sub_arr(idx)
 
         # calculate dual average: gradient
         g = ((t-1)/t).*g + (1/(t)).*dfunc(At,yt,w)

diff --git a/src/algorithms/sgd_alg.jl b/src/algorithms/sgd_alg.jl
@@ -12,11 +12,12 @@ function sgd_alg(dfunc::Function, X, Y, λ::Float64, k::Int, max_iter::Int, tole
 
     if ~check
         w = rand(d)
-        A = [X'; ones(1,N)]
+        sub_arr = (I) -> [sub(X,I,:) ones(k,1)]'
     else 
         total = length(X.nzval)
         w = sprand(d,1,total/(N*d))
-        A = [X'; sparse(ones(1,N))]
+        X = [X'; sparse(ones(1,N))]
+        sub_arr = (I) -> X[:,I]
     end
 
     if ~isempty(train_idx)
@@ -42,7 +43,7 @@ function sgd_alg(dfunc::Function, X, Y, λ::Float64, k::Int, max_iter::Int, tole
         w_prev = w
 
         yt = Y[idx]
-        At = A[:,idx]
+        At = sub_arr(idx)
 
         # do a gradient descent step
         η_t = 1/(λ*t)

diff --git a/src/algorithms/stochastic_rk_means.jl b/src/algorithms/stochastic_rk_means.jl
@@ -41,7 +41,7 @@ function stochastic_rk_means{A <: Algorithm}(X, ppc::RK_MEANS{A}, alg_params::Ve
     failed_mapping = false; t = 1; Y = ones(N)
 
     while true
-    	eval = pairwise(Euclidean(), X[train_idx,:]', w)
+    	eval = pairwise(Euclidean(), getindex(X,train_idx,:)', w)
     	(x,y) = findn(eval .== minimum(eval,2))
     	mappings = zeros(length(train_idx))
     	mappings[x] = y