diff --git a/weightmatrices/algos/ica.py b/weightmatrices/algos/ica.py
new file mode 100644
index 0000000..bf8c240
--- /dev/null
+++ b/weightmatrices/algos/ica.py
@@ -0,0 +1,16 @@
+
+import time
+from tqdm import tqdm
+
+from sklearn.decomposition import FastICA
+from weightmatrices.utils import utils
+
+def get_ica_trafo_matrix(data_loader, n_h):
+    transformer = FastICA(n_components=n_h)
+    
+
+def get_weightmatrices_ica(data_loader, n_h):
+    print("Creating weigth matrix for ICA for "+str(n_h)+" hidden neurons...")
+    s = data_loader.dataset.data.shape
+    n_in_features = s[-1]*s[-2]
+    assert n_h <= n_in_features, "Number of requested independent components higher than input dimensionality!"
diff --git a/weightmatrices/algos/sc.py b/weightmatrices/algos/sc.py
new file mode 100644
index 0000000..3fa98bd
--- /dev/null
+++ b/weightmatrices/algos/sc.py
@@ -0,0 +1,21 @@
+
+import time
+from tqdm import tqdm
+
+from sklearn.decomposition import dict_learning_online
+from weightmatrices.utils import utils
+
+
+
+def get_weightmatrices_sc(data_loader):
+    X = ?
+    out_tuple = dict_learning_online(X, n_components=2, *, alpha=1, n_iter=100,
+                             return_code=True, dict_init=None, callback=None,
+                             batch_size=3, verbose=False, shuffle=True,
+                             n_jobs=None, method='lars', iter_offset=0,
+                             random_state=None, return_inner_stats=False,
+                             inner_stats=None, return_n_iter=False,
+                             positive_dict=False, positive_code=False,
+                             method_max_iter=1000):
+    W = ?
+    return W
diff --git a/weightmatrices/utils/utils.py b/weightmatrices/utils/utils.py
index b726772..d145f9c 100644
--- a/weightmatrices/utils/utils.py
+++ b/weightmatrices/utils/utils.py
@@ -1,19 +1,24 @@
 
 import torch
 from torchvision import datasets, transforms
 from sklearn.preprocessing import normalize
 
 def load_data(dataset='EMNIST', batch_size=100000):
     if dataset=='EMNIST':
         trans = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,), (1.0,))])
         data = datasets.EMNIST('./datasets/', 'bymerge', transform=trans, download = True)
         data_loader = torch.utils.data.DataLoader(
                          dataset=data,
                          batch_size=batch_size,
                          shuffle=False,
                          drop_last=True)
 
     return data_loader
 
+def getbigdatamatrix(data_loader):
+    # Needed for ICA and SC and stuff...
+    # loop over data_loader and concatenation
+    return X
+
 def normalise_weightmatrix(W):
     return normalize(W)