m

Author: anddudkin

Examples/MNIST_example.py

@@ -7,21 +7,14 @@
 from NeuronModels import NeuronLifAdaptiveThresh
 import matplotlib.pyplot as plt
 
-n_neurons_out = 50
-n_neurons_in = 196
-n_train = 5000
-n_test = 800
-time = 350
-time_test = 200
-test = True
-plot = True
-
-conn = Connections(n_neurons_in, n_neurons_out, "all_to_all")
-conn.all_to_all_conn()
-conn.initialize_weights("normal")
-
-data_train = MNIST_train_test_14x14()[0]
-data_test = MNIST_train_test_14x14()[1]
+n_neurons_out = 50  # number of neurons in input layer
+n_neurons_in = 196  # number of output in input layer
+n_train = 5000  # number of images for training
+n_test = 800  # number of images for testing
+time = 350  # time of each image presentation during training
+time_test = 200  # time of each image presentation during testing
+test = True  # do testing or not
+plot = True  # plot graphics or not
 
 out_neurons = NeuronLifAdaptiveThresh(n_neurons_in,
                                       n_neurons_out,
@@ -32,7 +25,14 @@
                                       U_rest=0,
                                       refr_time=5,
                                       traces=True,
-                                      inh=True)
+                                      inh=True)  # activate literal inhibition
+
+conn = Connections(n_neurons_in, n_neurons_out, "all_to_all")
+conn.all_to_all_conn()
+conn.initialize_weights("normal")
+
+data_train = MNIST_train_test_14x14()[0]
+data_test = MNIST_train_test_14x14()[1]
 
 assig = MnistAssignment(n_neurons_out)
 
@@ -69,8 +69,6 @@
             assig.count_spikes_train(out_neurons.spikes, data_train[i][1])
             conn.update_w(out_neurons.spikes_trace_in, out_neurons.spikes_trace_out, out_neurons.spikes)
 
-fig.savefig('foo.png')
-fig1.savefig('fooo.png')
 assig.get_assigment()
 evall = MnistEvaluation(n_neurons_out)
 

__pycache__/NeuronModels.cpython-39.pyc

@@ -5,12 +5,12 @@
 from topology import Connections
 
 
-def plot_mnist():
+def plot_mnist_test():
     bb = MNIST_train_test_14x14()[0]
-    print(bb[0][0])
-    b1 = bb[0][0] * 0.80
-    f = torch.squeeze(encoding_to_spikes(b1, 10))
-    print(f)
+    spike_intensity = 1
+    b1 = bb[0][0] * spike_intensity
+    f = torch.squeeze(encoding_to_spikes(b1, 50))
+
     g = torch.zeros([14, 14], dtype=torch.float)
     plt.ion()
     fig = plt.figure(figsize=(4, 4))
@@ -23,16 +23,18 @@ def plot_mnist():
         plt.pause(0.0001)
 
     plt.ioff()
-    print(g)
+
     fig1 = plt.figure(figsize=(6, 6))
     ax3 = fig1.add_subplot(121)
     ax2 = fig1.add_subplot(122)
-    g = g / 10
+    g = g / 50
+    ax3.set_title('Image Sample')
+    ax2.set_title('Recreated image from spike train')
     ax3.imshow(torch.squeeze(bb[0][0]), cmap='gray', vmin=0, vmax=1)
     ax2.imshow(g, cmap='gray', vmin=0, vmax=1)
     plt.show()
 
-
+plot_mnist_test()
 def if_neuron_test():
     """ Visualize how single IF neuron works.
     """
@@ -165,4 +167,3 @@ def lif_thresh_neuron_test():
     plt.axhline(y=lif_test.U_tr, color='k')
     plt.show()
 
-lif_thresh_neuron_test()