from future import division import numpy as np import matplotlib.pyplot as plt import matplotlib.gridspec as gridspec from sklearn import datasets from neupy import algorithms, environment
!pip install neupy
environment.reproducible() GRID_HEIGHT = 40 GRID_WIDTH = 40 digits = datasets.load_digits() data = digits.data sofm = algorithms.SOFM( n_inputs=64, features_grid=(GRID_HEIGHT, GRID_WIDTH), # Learning radius defines area within which we find # winning neuron neighbours. The higher the value # the more values we will be updated after each iteration. learning_radius=5, # Every 20 epochs learning radius will be reduced by 1. reduce_radius_after=20, step=0.5, std=1, shuffle_data=True, verbose=True, ) Output: Main information [ALGORITHM] SOFM [OPTION] distance = ['euclid', 'neg_euclid_distance'] [OPTION] features_grid = [40, 40] [OPTION] grid_type = ['rectangle', 'find_neighbours_on_rect_grid', 'find_step_scaler_on_rect_grid'] [OPTION] learning_radius = 5 [OPTION] n_inputs = 64 [OPTION] n_outputs = None [OPTION] reduce_radius_after = 20 [OPTION] reduce_std_after = 100 [OPTION] reduce_step_after = 100 [OPTION] show_epoch = 1 [OPTION] shuffle_data = True [OPTION] signals = None [OPTION] std = 1 [OPTION] step = 0.5 [OPTION] verbose = True [OPTION] weight = Normal(mean=0, std=0.01)
import matplotlib.inline for row_id in range(GRID_HEIGHT): print("Progress: {:.2%}".format(row_id / GRID_HEIGHT)) for col_id in range(GRID_WIDTH): index = row_id * GRID_HEIGHT + col_id clustered_samples = data[clusters == index] if len(clustered_samples) > 0: # We take the first sample, but it can be any # sample from this cluster sample = clustered_samples[0] else: # If we don't have samples in cluster then # it means that there is a gap in space sample = np.zeros(64) plt.subplot(grid[index]) plt.imshow(sample.reshape((8, 8)), cmap='Greys') plt.axis('off') print("Visualization has been built succesfully") plt.show()
