implementing normal.py.cs API log_prob() for NB classifier prediction

Author: PppBr

src/TensorFlowNET.Core/Operations/Distributions/distribution.py.cs

@@ -27,6 +27,37 @@ public class Distribution : _BaseDistribution
         public List<Tensor> _graph_parents  {get;set;}
         public string _name  {get;set;}
 
+
+        /// <summary>
+        /// Log probability density/mass function.
+        /// </summary>
+        /// <param name="value"> `Tensor`.</param>
+        /// <param name="name"> Python `str` prepended to names of ops created by this function.</param>
+        /// <returns>log_prob: a `Tensor` of shape `sample_shape(x) + self.batch_shape` with values of type `self.dtype`.</returns>
+
+        /*
+        public Tensor log_prob(Tensor value, string name = "log_prob")
+        {
+            return _call_log_prob(value, name);
+        }
+
+        private Tensor _call_log_prob (Tensor value, string name)
+        {
+            with(ops.name_scope(name, "moments", new { value }), scope =>
+            {
+                value = _convert_to_tensor(value, "value", _dtype);
+            });
+
+            throw new NotImplementedException();
+
+        }
+
+        private Tensor _convert_to_tensor(Tensor value, string name = null, TF_DataType preferred_dtype)
+        {
+            throw new NotImplementedException();
+        }
+        */
+
         /// <summary>
         /// Constructs the `Distribution'     
         /// **This is a private method for subclass use.**
@@ -47,7 +78,7 @@ public class Distribution : _BaseDistribution
         /// <param name = "name"> Name prefixed to Ops created by this class. Default: subclass name.</param>
         /// <returns> Two `Tensor` objects: `mean` and `variance`.</returns>
 
-        /* 
+        /*
         private Distribution (
                 TF_DataType dtype,
                 ReparameterizationType reparameterization_type,
@@ -66,6 +97,10 @@ private Distribution (
                     this._name = name;
                 }
         */
+
+
+
+
     }
 
     /// <summary>

src/TensorFlowNET.Core/Operations/Distributions/normal.py.cs

@@ -78,5 +78,19 @@ public Tensor _batch_shape()
             return array_ops.broadcast_static_shape(new Tensor(_loc.shape), new Tensor(_scale.shape));
         }
 
+        private Tensor _log_prob(Tensor x)
+        {
+            return _log_unnormalized_prob(_z(x));
+        }
+
+        private Tensor _log_unnormalized_prob (Tensor x)
+        {
+            return -0.5 * math_ops.square(_z(x));
+        }
+
+        private Tensor _z (Tensor x)
+        {
+            return (x - this._loc) / this._scale;
+        }
     }
 }

src/TensorFlowNET.Core/Operations/math_ops.py.cs

@@ -55,6 +55,11 @@ public static Tensor square_difference(Tensor x, Tensor y, string name = null)
             return m;
         }
 
+        public static Tensor square(Tensor x, string name = null)
+        {
+            throw new NotImplementedException();
+        }
+
         /// <summary>
         /// Helper function for reduction ops.
         /// </summary>

test/TensorFlowNET.Examples/NaiveBayesClassifier.cs

@@ -15,13 +15,10 @@ public class NaiveBayesClassifier : Python, IExample
         public void Run()
         {
             np.array<float>(1.0f, 1.0f);
-            // var X = np.array<float>(np.array<float>(1.0f, 1.0f), np.array<float>(2.0f, 2.0f), np.array<float>(1.0f, -1.0f), np.array<float>(2.0f, -2.0f), np.array<float>(-1.0f, -1.0f), np.array<float>(-1.0f, 1.0f),);
-            // var X = np.array<float[]>(new float[][] { new float[] { 1.0f, 1.0f}, new float[] { 2.0f, 2.0f }, new float[] { -1.0f, -1.0f }, new float[] { -2.0f, -2.0f }, new float[] { 1.0f, -1.0f }, new float[] { 2.0f, -2.0f }, });
             var X = np.array<float>(new float[][] { new float[] { 1.0f, 1.0f }, new float[] { 2.0f, 2.0f }, new float[] { -1.0f, -1.0f }, new float[] { -2.0f, -2.0f }, new float[] { 1.0f, -1.0f }, new float[] { 2.0f, -2.0f }, });
             var y = np.array<int>(0,0,1,1,2,2);
             fit(X, y);
             // Create a regular grid and classify each point 
-            
         }
 
         public void fit(NDArray X, NDArray y)