API tools faq
paste
Advertisement
cimona

Validacija

cimona
Sep 13th, 2018
359
0
Never
Add comment
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 8.74 KB | None | 0 0
raw download clone embed print report
  1. from math import exp
  2. from random import seed
  3. from random import random
  4.  
  5. # Initialize a network
  6. # Put fixed weigths to 0.5 on code.finki.ukim.mk if there is a problem with random()
  7. def initialize_network(n_inputs, n_hidden, n_outputs):
  8. network = list()
  9. hidden_layer = [{'weights':[random() for i in range(n_inputs + 1)]} for i in range(n_hidden)]
  10. network.append(hidden_layer)
  11. output_layer = [{'weights':[random() for i in range(n_hidden + 1)]} for i in range(n_outputs)]
  12. network.append(output_layer)
  13. return network
  14.  
  15.  
  16. # Calculate neuron activation for an input
  17. def activate(weights, inputs):
  18. activation = weights[-1]
  19. for i in range(len(weights)-1):
  20. activation += weights[i] * inputs[i]
  21. return activation
  22.  
  23. # Transfer neuron activation
  24. def transfer(activation):
  25. return 1.0 / (1.0 + exp(-activation))
  26.  
  27. # Forward propagate input to a network output
  28. def forward_propagate(network, row):
  29. inputs = row
  30. for layer in network:
  31. new_inputs = []
  32. for neuron in layer:
  33. activation = activate(neuron['weights'], inputs)
  34. neuron['output'] = transfer(activation)
  35. new_inputs.append(neuron['output'])
  36. inputs = new_inputs
  37. return inputs
  38.  
  39. # Calculate the derivative of an neuron output
  40. def transfer_derivative(output):
  41. return output * (1.0 - output)
  42.  
  43. # Backpropagate error and store in neurons
  44. def backward_propagate_error(network, expected):
  45. for i in reversed(range(len(network))):
  46. layer = network[i]
  47. errors = list()
  48. if i != len(network)-1:
  49. for j in range(len(layer)):
  50. error = 0.0
  51. for neuron in network[i + 1]:
  52. error += (neuron['weights'][j] * neuron['delta'])
  53. errors.append(error)
  54. else:
  55. for j in range(len(layer)):
  56. neuron = layer[j]
  57. errors.append(expected[j] - neuron['output'])
  58. for j in range(len(layer)):
  59. neuron = layer[j]
  60. neuron['delta'] = errors[j] * transfer_derivative(neuron['output'])
  61.  
  62. # Update network weights with error
  63. def update_weights(network, row, l_rate):
  64. for i in range(len(network)):
  65. inputs = row[:-1]
  66. if i != 0:
  67. inputs = [neuron['output'] for neuron in network[i - 1]]
  68. for neuron in network[i]:
  69. for j in range(len(inputs)):
  70. neuron['weights'][j] += l_rate * neuron['delta'] * inputs[j]
  71. neuron['weights'][-1] += l_rate * neuron['delta']
  72.  
  73. # Train a network for a fixed number of epochs
  74. def train_network(network, train, l_rate, n_epoch, n_outputs):
  75. for epoch in range(n_epoch):
  76. sum_error = 0
  77. for row in train:
  78. outputs = forward_propagate(network, row)
  79. expected = [0 for i in range(n_outputs)]
  80. expected[row[-1]] = 1
  81. sum_error += sum([(expected[i]-outputs[i])**2 for i in range(len(expected))])
  82. backward_propagate_error(network, expected)
  83. update_weights(network, row, l_rate)
  84.  
  85.  
  86. # Test training backprop algorithm
  87.  
  88.  
  89.  
  90.  
  91.  
  92.  
  93. dataset = [
  94. [6.3, 2.9, 5.6, 1.8, 0],
  95. [6.5, 3.0, 5.8, 2.2, 0],
  96. [7.6, 3.0, 6.6, 2.1, 0],
  97. [4.9, 2.5, 4.5, 1.7, 0],
  98. [7.3, 2.9, 6.3, 1.8, 0],
  99. [6.7, 2.5, 5.8, 1.8, 0],
  100. [7.2, 3.6, 6.1, 2.5, 0],
  101. [6.5, 3.2, 5.1, 2.0, 0],
  102. [6.4, 2.7, 5.3, 1.9, 0],
  103. [6.8, 3.0, 5.5, 2.1, 0],
  104. [5.7, 2.5, 5.0, 2.0, 0],
  105. [5.8, 2.8, 5.1, 2.4, 0],
  106. [6.4, 3.2, 5.3, 2.3, 0],
  107. [6.5, 3.0, 5.5, 1.8, 0],
  108. [7.7, 3.8, 6.7, 2.2, 0],
  109. [7.7, 2.6, 6.9, 2.3, 0],
  110. [6.0, 2.2, 5.0, 1.5, 0],
  111. [6.9, 3.2, 5.7, 2.3, 0],
  112. [5.6, 2.8, 4.9, 2.0, 0],
  113. [7.7, 2.8, 6.7, 2.0, 0],
  114. [6.3, 2.7, 4.9, 1.8, 0],
  115. [6.7, 3.3, 5.7, 2.1, 0],
  116. [7.2, 3.2, 6.0, 1.8, 0],
  117. [6.2, 2.8, 4.8, 1.8, 0],
  118. [6.1, 3.0, 4.9, 1.8, 0],
  119. [6.4, 2.8, 5.6, 2.1, 0],
  120. [7.2, 3.0, 5.8, 1.6, 0],
  121. [7.4, 2.8, 6.1, 1.9, 0],
  122. [7.9, 3.8, 6.4, 2.0, 0],
  123. [6.4, 2.8, 5.6, 2.2, 0],
  124. [6.3, 2.8, 5.1, 1.5, 0],
  125. [6.1, 2.6, 5.6, 1.4, 0],
  126. [7.7, 3.0, 6.1, 2.3, 0],
  127. [6.3, 3.4, 5.6, 2.4, 0],
  128. [5.1, 3.5, 1.4, 0.2, 1],
  129. [4.9, 3.0, 1.4, 0.2, 1],
  130. [4.7, 3.2, 1.3, 0.2, 1],
  131. [4.6, 3.1, 1.5, 0.2, 1],
  132. [5.0, 3.6, 1.4, 0.2, 1],
  133. [5.4, 3.9, 1.7, 0.4, 1],
  134. [4.6, 3.4, 1.4, 0.3, 1],
  135. [5.0, 3.4, 1.5, 0.2, 1],
  136. [4.4, 2.9, 1.4, 0.2, 1],
  137. [4.9, 3.1, 1.5, 0.1, 1],
  138. [5.4, 3.7, 1.5, 0.2, 1],
  139. [4.8, 3.4, 1.6, 0.2, 1],
  140. [4.8, 3.0, 1.4, 0.1, 1],
  141. [4.3, 3.0, 1.1, 0.1, 1],
  142. [5.8, 4.0, 1.2, 0.2, 1],
  143. [5.7, 4.4, 1.5, 0.4, 1],
  144. [5.4, 3.9, 1.3, 0.4, 1],
  145. [5.1, 3.5, 1.4, 0.3, 1],
  146. [5.7, 3.8, 1.7, 0.3, 1],
  147. [5.1, 3.8, 1.5, 0.3, 1],
  148. [5.4, 3.4, 1.7, 0.2, 1],
  149. [5.1, 3.7, 1.5, 0.4, 1],
  150. [4.6, 3.6, 1.0, 0.2, 1],
  151. [5.1, 3.3, 1.7, 0.5, 1],
  152. [4.8, 3.4, 1.9, 0.2, 1],
  153. [5.0, 3.0, 1.6, 0.2, 1],
  154. [5.0, 3.4, 1.6, 0.4, 1],
  155. [5.2, 3.5, 1.5, 0.2, 1],
  156. [5.2, 3.4, 1.4, 0.2, 1],
  157. [5.5, 2.3, 4.0, 1.3, 2],
  158. [6.5, 2.8, 4.6, 1.5, 2],
  159. [5.7, 2.8, 4.5, 1.3, 2],
  160. [6.3, 3.3, 4.7, 1.6, 2],
  161. [4.9, 2.4, 3.3, 1.0, 2],
  162. [6.6, 2.9, 4.6, 1.3, 2],
  163. [5.2, 2.7, 3.9, 1.4, 2],
  164. [5.0, 2.0, 3.5, 1.0, 2],
  165. [5.9, 3.0, 4.2, 1.5, 2],
  166. [6.0, 2.2, 4.0, 1.0, 2],
  167. [6.1, 2.9, 4.7, 1.4, 2],
  168. [5.6, 2.9, 3.6, 1.3, 2],
  169. [6.7, 3.1, 4.4, 1.4, 2],
  170. [5.6, 3.0, 4.5, 1.5, 2],
  171. [5.8, 2.7, 4.1, 1.0, 2],
  172. [6.2, 2.2, 4.5, 1.5, 2],
  173. [5.6, 2.5, 3.9, 1.1, 2],
  174. [5.9, 3.2, 4.8, 1.8, 2],
  175. [6.1, 2.8, 4.0, 1.3, 2],
  176. [6.3, 2.5, 4.9, 1.5, 2],
  177. [6.1, 2.8, 4.7, 1.2, 2],
  178. [6.4, 2.9, 4.3, 1.3, 2],
  179. [6.6, 3.0, 4.4, 1.4, 2],
  180. [6.8, 2.8, 4.8, 1.4, 2],
  181. [6.7, 3.0, 5.0, 1.7, 2],
  182. [6.0, 2.9, 4.5, 1.5, 2],
  183. [5.7, 2.6, 3.5, 1.0, 2],
  184. [5.5, 2.4, 3.8, 1.1, 2],
  185. [5.4, 3.0, 4.5, 1.5, 2],
  186. [6.0, 3.4, 4.5, 1.6, 2],
  187. [6.7, 3.1, 4.7, 1.5, 2],
  188. [6.3, 2.3, 4.4, 1.3, 2],
  189. [5.6, 3.0, 4.1, 1.3, 2],
  190. [5.5, 2.5, 4.0, 1.3, 2],
  191. [5.5, 2.6, 4.4, 1.2, 2],
  192. [6.1, 3.0, 4.6, 1.4, 2],
  193. [5.8, 2.6, 4.0, 1.2, 2],
  194. [5.0, 2.3, 3.3, 1.0, 2],
  195. [5.6, 2.7, 4.2, 1.3, 2],
  196. [5.7, 3.0, 4.2, 1.2, 2],
  197. [5.7, 2.9, 4.2, 1.3, 2],
  198. [6.2, 2.9, 4.3, 1.3, 2],
  199. [5.1, 2.5, 3.0, 1.1, 2],
  200. [5.7, 2.8, 4.1, 1.3, 2],
  201. [6.4, 3.1, 5.5, 1.8, 0],
  202. [6.0, 3.0, 4.8, 1.8, 0],
  203. [6.9, 3.1, 5.4, 2.1, 0],
  204. [6.8, 3.2, 5.9, 2.3, 0],
  205. [6.7, 3.3, 5.7, 2.5, 0],
  206. [6.7, 3.0, 5.2, 2.3, 0],
  207. [6.3, 2.5, 5.0, 1.9, 0],
  208. [6.5, 3.0, 5.2, 2.0, 0],
  209. [6.2, 3.4, 5.4, 2.3, 0],
  210. [4.7, 3.2, 1.6, 0.2, 1],
  211. [4.8, 3.1, 1.6, 0.2, 1],
  212. [5.4, 3.4, 1.5, 0.4, 1],
  213. [5.2, 4.1, 1.5, 0.1, 1],
  214. [5.5, 4.2, 1.4, 0.2, 1],
  215. [4.9, 3.1, 1.5, 0.2, 1],
  216. [5.0, 3.2, 1.2, 0.2, 1],
  217. [5.5, 3.5, 1.3, 0.2, 1],
  218. [4.9, 3.6, 1.4, 0.1, 1],
  219. [4.4, 3.0, 1.3, 0.2, 1],
  220. [5.1, 3.4, 1.5, 0.2, 1],
  221. [5.0, 3.5, 1.3, 0.3, 1],
  222. [4.5, 2.3, 1.3, 0.3, 1],
  223. [4.4, 3.2, 1.3, 0.2, 1],
  224. [5.0, 3.5, 1.6, 0.6, 1],
  225. [5.9, 3.0, 5.1, 1.8, 0],
  226. [5.1, 3.8, 1.9, 0.4, 1],
  227. [4.8, 3.0, 1.4, 0.3, 1],
  228. [5.1, 3.8, 1.6, 0.2, 1],
  229. [5.5, 2.4, 3.7, 1.0, 2],
  230. [5.8, 2.7, 3.9, 1.2, 2],
  231. [6.0, 2.7, 5.1, 1.6, 2],
  232. [6.7, 3.1, 5.6, 2.4, 0],
  233. [6.9, 3.1, 5.1, 2.3, 0],
  234. [5.8, 2.7, 5.1, 1.9, 0],
  235. ]
  236.  
  237. def predict(network,row):
  238. outputs = forward_propagate(network,row)
  239. return outputs.index(max(outputs))
  240.  
  241. def getScore(network,data):
  242. score = 0
  243. for row in data:
  244. prediction = predict(network,row)
  245. if (prediction == row[-1]):
  246. score += 1
  247. return score
  248.  
  249. def getDataset(dataset):
  250. list = []
  251. validation = []
  252. for i in range(0, len(dataset) - 10):
  253. list.append(dataset[i])
  254.  
  255. for i in range(len(dataset) - 10, len(dataset)):
  256. validation.append(dataset[i])
  257.  
  258. return (list,validation)
  259. if __name__ == "__main__":
  260. # ne menuvaj
  261. seed(1)
  262.  
  263. att1 = input()
  264. att2 = input()
  265. att3 = input()
  266. att4 = input()
  267. planttype = input()
  268. testCase = [att1, att2, att3, att4, planttype]
  269. data = getDataset(dataset)[0]
  270. validation = getDataset(dataset)[1]
  271.  
  272. n_inputs=len(data[0]) -1
  273. n_outputs=len(set(row[-1] for row in data))
  274.  
  275. network1=initialize_network(n_inputs,3,n_outputs)
  276. network2=initialize_network(n_inputs,3,n_outputs)
  277. network3=initialize_network(n_inputs,3,n_outputs)
  278.  
  279. train_network(network1,data,0.3,20,n_outputs)
  280. train_network(network2, data, 0.5, 20, n_outputs)
  281. train_network(network3, data, 0.7, 20, n_outputs)
  282.  
  283. score_network1 = getScore(network1,validation)
  284. score_network2 = getScore(network2,validation)
  285. score_network3 = getScore(network3,validation)
  286.  
  287. MAX = max(score_network1,score_network2,score_network3)
  288.  
  289. if (MAX == score_network1):
  290.  
  291. prediction = predict(network1, testCase)
  292. print(prediction)
  293. elif (MAX == score_network2):
  294.  
  295. prediction = predict(network2, testCase)
  296. print(prediction)
  297. else :
  298.  
  299. prediction = predict(network3, testCase)
  300. print(prediction)
  301. # vasiot kod ovde
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
Public Pastes
Advertisement
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy.  OK, I Understand
Not a member of Pastebin yet?
Sign Up, it unlocks many cool features!