Accuracy pendant training

sobek/network.py

@@ -60,14 +60,14 @@ class network:
 
 
 
-    def train(self, inputs, desiredOutputs, learningRate, batchSize, epochs=1, visualize=False):
+    def train(self, inputs, desiredOutputs, learningRate, batchSize, epochs=1, accuracyInputs=False, accuracyDesiredOutputs=False, visualize=False):
         if (type(inputs) != list or type(desiredOutputs) != list):
             raise TypeError("The inputs and desired outputs must be lists of numpy arrays !")
         if (len(inputs) != len(desiredOutputs)):
             raise ValueError("The inputs and desired outputs lists must have the same amount of data ! " + str(len(inputs)) + " != " + str(len(desiredOutputs)))
         if (len(inputs) == 0):
             raise ValueError("The list is empty !")
-        if (visualize == False):
+        if (visualize != False):
             if (self.__inputLayerSize != 2):
                 raise ValueError("Visualization is only possible for 2 inputs networks")
             if (len(self.weights[-1]) != 1):
@@ -126,6 +126,14 @@ class network:
                     errorSumsBiases[layerNumber] = np.multiply(errorSumsBiases[layerNumber], -(learningRate/len(inputBatch)))
                     self.biases[layerNumber] = np.add(self.biases[layerNumber], errorSumsBiases[layerNumber])
 
+            '''errorSum = 0
+            for layer in errorSumsBiases:
+                errorSum += np.sum(np.abs(layer))
+            print(errorSum)'''
+
+            if (accuracyInputs and accuracyDesiredOutputs):
+                print(self.accuracy(accuracyInputs, accuracyDesiredOutputs))
+
         if (visualize):
             ani = animation.ArtistAnimation(fig, vizualisationData, interval=100)
             plt.show()

tests/MNISTDrawingPrediction.py

@@ -14,16 +14,19 @@ class Sketchpad(tkinter.Canvas):
         self.bind("<B1-Motion>", self.add_line)
         self.PILImage = Image.new("F", (560, 560), 100)
         self.draw = ImageDraw.Draw(self.PILImage)
-        self.MNISTNN = network.networkFromFile("MNIST30epoch")
+        self.MNISTNN = network.networkFromFile("MNISTtest2")
         self.predictionLabel = predictionLabel
 
     def add_line(self, event):
-        self.create_oval((event.x+32, event.y+32, event.x-32, event.y-32), fill="black")
-        self.draw.ellipse([event.x-32, event.y-32, event.x+32, event.y+32], fill="black")
+        self.create_oval((event.x+24, event.y+24, event.x-24, event.y-24), fill="black")
+        self.draw.ellipse([event.x-24, event.y-24, event.x+24, event.y+24], fill="black")
         smallerImage = self.PILImage.reduce(20)
         imageAsArray = np.array(smallerImage.getdata())
         imageAsArray = (100 - imageAsArray)/100
-        self.predictionLabel['text'] =  ( "Predicted number : "  + str(np.argmax(self.MNISTNN.process(imageAsArray))))
+        predictionArray = self.MNISTNN.process(imageAsArray)
+        print(predictionArray)
+        prediction = np.argmax(predictionArray)
+        self.predictionLabel['text'] =  ( "Predicted number : "  + str(prediction) + " with confidence : " + str(predictionArray[prediction]))
 
     def test(self, event):
         self.PILImage = Image.new("F", (560, 560), 100)

tests/MNISTLearning.py

@@ -27,7 +27,20 @@ for label in tempTrainLabels:
     trainLabels.append(np.zeros(10))
     trainLabels[-1][label] = 1.0
 
-myNetwork = network(784, 30, 10)
+tempAccuracyImages = getData("./MNIST/t10k-images-idx3-ubyte.gz")[0x10:].reshape((-1, 784)).tolist()
+accuracyImages = []
+for image in tempAccuracyImages:
+    for pixel in range(784):
+        if image[pixel] !=0:
+            image[pixel] = image[pixel]/256
+    accuracyImages.append(np.array(image, dtype=np.float64))
+tempAccuracyLabels = getData("./MNIST/t10k-labels-idx1-ubyte.gz")[8:]
+accuracyLabels = []
+for label in tempAccuracyLabels:
+    accuracyLabels.append(np.zeros(10))
+    accuracyLabels[-1][label] = 1.0
+
+myNetwork = network(784, 32, 10)
 
 learningRate = 3.0
 
@@ -48,7 +61,7 @@ for i in range(1):
             print(batchEnd)
 """
 
-myNetwork.train(trainImages, trainLabels, learningRate, 10, 30)
+myNetwork.train(trainImages, trainLabels, learningRate, batchSize=10, epochs=30, accuracyInputs=accuracyImages, accuracyDesiredOutputs=accuracyLabels)
 
 endTime = time.perf_counter()
 
@@ -57,4 +70,4 @@ print("Learning time : " + str(endTime - startTime))
 print(trainLabels[121])
 print(myNetwork.process(trainImages[121]))
 
-myNetwork.saveToFile("MNIST30epoch")
+myNetwork.saveToFile("MNISTtest3")

tests/testLearningNAND.py

@@ -35,7 +35,7 @@ for i in range(nbRep):
 
     startTime = time.perf_counter()
 
-    myNetwork.train(test, result, learningRate, len(test), 10000, visualize=False)
+    myNetwork.train(test, result, learningRate, len(test), 10000, visualize=True)
 
     endTime = time.perf_counter()
     learningTime += endTime - startTime