vaniall cpse resolved #34

README.md

@@ -43,6 +43,24 @@ pip install -upgrade git+https://github.com/msuzen/bristol.git
 ## Documentation
 ### Complexity of a deep learning model: cPSE
 
+### Vanilla case
+
+In the vanilla case a list of matrices that are representative of 
+ordered set of weight matrices can be used to compute cPSE over
+layers. As an examples: 
+
+```python
+from bristol import cPSE
+import numpy as np
+np.random.seed(42)
+matrices = [np.random.normal(size=(64,64)) for _ in range(10)]
+(d_layers, cpse) = cPSE.cpse_measure_vanilla(matrices)
+```
+Even for set of Gaussian matrices, d_layers decrease.
+Note that different layer types should be converted to a matrix
+format, i.e., CNNs to 2D matrices. See the main paper.
+
+### For torch models
 You need to put your model as pretrained model format of PyTorch. An example for vgg, 
 and use `cPSE.cpse_measure` function simply:
 

bristol/cPSE.py

@@ -161,4 +161,19 @@ def cpse_measure(pmodel):
     d_layers = d_layers_pse(eset_per)
     return d_layers, np.mean(np.log10(d_layers))
 
+def cpse_measure_vanilla(matrices):
+    """
+
+    Given list of weight matrices.
+    pse on layers and mean log pse Cascading PSE
+    (d_layers, cpse) : d_layers vector and real number cpse
+
+    np.random.seed(42)
+    matrices = [np.random.normal(size=(64,64)) for _ in range(10)]
+    (d_layers, cpse) = cPSE.cpse_measure_vanilla(matrices)
 
+    """
+    eset = get_eigenvals_layer_matrix_set(matrices)
+    eset_per = eigenvals_set_to_periodic(eset)
+    d_layers = d_layers_pse(eset_per)
+    return d_layers, np.mean(np.log10(d_layers))

bristol/version.py

@@ -1 +1 @@
-__version__='0.2.11'
+__version__='0.2.12'