Tue May 7 10:29:45 2019 paraheat_gaussian_parameter: python version: 3.6.8 keras version: 2.2.4 Neural network to solve a multivariate regression problem. Estimate the parameter vc used in a Gaussian diffusivity given vs, 50 samples of the resulting heat distribution Data of many records of vc and vs is available. The data is read from an external file. Read data from paraheat_gaussian_parameter_data.txt Data contains 500 records with 55 features. Training data uses 475 records with 50 features and 1 targets. Test data uses 25 records with 50 features and 1 targets. train_data[0,0:10]: [ 9.6265467 13.613504 27.642524 23.320682 25.420298 20.927116 12.464001 6.2774213 23.433233 24.045476 ] train_targets[0]: 2.8977486 test_data[0,0:10]: [ 9.629448 13.617584 27.651117 23.327865 25.428162 20.933537 12.46775 6.2793045 23.440456 24.052893 ] test_targets[0]: 2.8966315 Using 200 nodes per layer _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= dense_1 (Dense) (None, 200) 10200 _________________________________________________________________ dense_2 (Dense) (None, 200) 40200 _________________________________________________________________ dense_3 (Dense) (None, 200) 40200 _________________________________________________________________ dense_4 (Dense) (None, 200) 40200 _________________________________________________________________ dense_5 (Dense) (None, 200) 40200 _________________________________________________________________ dense_6 (Dense) (None, 200) 40200 _________________________________________________________________ dense_7 (Dense) (None, 200) 40200 _________________________________________________________________ dense_8 (Dense) (None, 1) 201 ================================================================= Total params: 251,601 Trainable params: 251,601 Non-trainable params: 0 _________________________________________________________________ Training: Train on 380 samples, validate on 95 samples Epoch 1/30 32/380 [=>............................] - ETA: 3s - loss: 4.6541 - mean_squared_error: 4.6541 380/380 [==============================] - 0s 895us/step - loss: 7.9935 - mean_squared_error: 7.9935 - val_loss: 5.1423 - val_mean_squared_error: 5.1423 Epoch 2/30 32/380 [=>............................] - ETA: 0s - loss: 6.7038 - mean_squared_error: 6.7038 380/380 [==============================] - 0s 76us/step - loss: 5.1115 - mean_squared_error: 5.1115 - val_loss: 4.8968 - val_mean_squared_error: 4.8968 Epoch 3/30 32/380 [=>............................] - ETA: 0s - loss: 4.5375 - mean_squared_error: 4.5375 380/380 [==============================] - 0s 75us/step - loss: 5.3298 - mean_squared_error: 5.3298 - val_loss: 5.0117 - val_mean_squared_error: 5.0117 Epoch 4/30 32/380 [=>............................] - ETA: 0s - loss: 5.2104 - mean_squared_error: 5.2104 380/380 [==============================] - 0s 78us/step - loss: 4.5679 - mean_squared_error: 4.5679 - val_loss: 4.4033 - val_mean_squared_error: 4.4033 Epoch 5/30 32/380 [=>............................] - ETA: 0s - loss: 3.5708 - mean_squared_error: 3.5708 380/380 [==============================] - 0s 75us/step - loss: 4.3936 - mean_squared_error: 4.3936 - val_loss: 4.1381 - val_mean_squared_error: 4.1381 Epoch 6/30 32/380 [=>............................] - ETA: 0s - loss: 3.7348 - mean_squared_error: 3.7348 380/380 [==============================] - 0s 73us/step - loss: 4.1137 - mean_squared_error: 4.1137 - val_loss: 3.7512 - val_mean_squared_error: 3.7512 Epoch 7/30 32/380 [=>............................] - ETA: 0s - loss: 4.6712 - mean_squared_error: 4.6712 380/380 [==============================] - 0s 75us/step - loss: 3.5547 - mean_squared_error: 3.5547 - val_loss: 3.1399 - val_mean_squared_error: 3.1399 Epoch 8/30 32/380 [=>............................] - ETA: 0s - loss: 2.1384 - mean_squared_error: 2.1384 380/380 [==============================] - 0s 73us/step - loss: 3.0003 - mean_squared_error: 3.0003 - val_loss: 2.3990 - val_mean_squared_error: 2.3990 Epoch 9/30 32/380 [=>............................] - ETA: 0s - loss: 2.0532 - mean_squared_error: 2.0532 380/380 [==============================] - 0s 74us/step - loss: 3.6533 - mean_squared_error: 3.6533 - val_loss: 4.3749 - val_mean_squared_error: 4.3749 Epoch 10/30 32/380 [=>............................] - ETA: 0s - loss: 5.3775 - mean_squared_error: 5.3775 380/380 [==============================] - 0s 73us/step - loss: 3.6122 - mean_squared_error: 3.6122 - val_loss: 2.8217 - val_mean_squared_error: 2.8217 Epoch 11/30 32/380 [=>............................] - ETA: 0s - loss: 2.0232 - mean_squared_error: 2.0232 380/380 [==============================] - 0s 73us/step - loss: 2.2164 - mean_squared_error: 2.2164 - val_loss: 1.6419 - val_mean_squared_error: 1.6419 Epoch 12/30 32/380 [=>............................] - ETA: 0s - loss: 0.9881 - mean_squared_error: 0.9881 380/380 [==============================] - 0s 74us/step - loss: 1.6402 - mean_squared_error: 1.6402 - val_loss: 3.0404 - val_mean_squared_error: 3.0404 Epoch 13/30 32/380 [=>............................] - ETA: 0s - loss: 2.9118 - mean_squared_error: 2.9118 380/380 [==============================] - 0s 74us/step - loss: 1.7395 - mean_squared_error: 1.7395 - val_loss: 1.5117 - val_mean_squared_error: 1.5117 Epoch 14/30 32/380 [=>............................] - ETA: 0s - loss: 1.2912 - mean_squared_error: 1.2912 380/380 [==============================] - 0s 75us/step - loss: 0.7727 - mean_squared_error: 0.7727 - val_loss: 0.6344 - val_mean_squared_error: 0.6344 Epoch 15/30 32/380 [=>............................] - ETA: 0s - loss: 0.6375 - mean_squared_error: 0.6375 380/380 [==============================] - 0s 77us/step - loss: 0.5113 - mean_squared_error: 0.5113 - val_loss: 0.2421 - val_mean_squared_error: 0.2421 Epoch 16/30 32/380 [=>............................] - ETA: 0s - loss: 0.2664 - mean_squared_error: 0.2664 380/380 [==============================] - 0s 73us/step - loss: 0.1767 - mean_squared_error: 0.1767 - val_loss: 0.1081 - val_mean_squared_error: 0.1081 Epoch 17/30 32/380 [=>............................] - ETA: 0s - loss: 0.1390 - mean_squared_error: 0.1390 380/380 [==============================] - 0s 73us/step - loss: 0.0886 - mean_squared_error: 0.0886 - val_loss: 0.1527 - val_mean_squared_error: 0.1527 Epoch 18/30 32/380 [=>............................] - ETA: 0s - loss: 0.1747 - mean_squared_error: 0.1747 380/380 [==============================] - 0s 73us/step - loss: 0.1083 - mean_squared_error: 0.1083 - val_loss: 0.1246 - val_mean_squared_error: 0.1246 Epoch 19/30 32/380 [=>............................] - ETA: 0s - loss: 0.1219 - mean_squared_error: 0.1219 380/380 [==============================] - 0s 76us/step - loss: 0.2904 - mean_squared_error: 0.2904 - val_loss: 0.3653 - val_mean_squared_error: 0.3653 Epoch 20/30 32/380 [=>............................] - ETA: 0s - loss: 0.2749 - mean_squared_error: 0.2749 380/380 [==============================] - 0s 73us/step - loss: 0.1322 - mean_squared_error: 0.1322 - val_loss: 0.0329 - val_mean_squared_error: 0.0329 Epoch 21/30 32/380 [=>............................] - ETA: 0s - loss: 0.0444 - mean_squared_error: 0.0444 380/380 [==============================] - 0s 73us/step - loss: 0.0328 - mean_squared_error: 0.0328 - val_loss: 0.0343 - val_mean_squared_error: 0.0343 Epoch 22/30 32/380 [=>............................] - ETA: 0s - loss: 0.0428 - mean_squared_error: 0.0428 380/380 [==============================] - 0s 73us/step - loss: 0.0290 - mean_squared_error: 0.0290 - val_loss: 0.0143 - val_mean_squared_error: 0.0143 Epoch 23/30 32/380 [=>............................] - ETA: 0s - loss: 0.0142 - mean_squared_error: 0.0142 380/380 [==============================] - 0s 74us/step - loss: 0.0336 - mean_squared_error: 0.0336 - val_loss: 0.0462 - val_mean_squared_error: 0.0462 Epoch 24/30 32/380 [=>............................] - ETA: 0s - loss: 0.0449 - mean_squared_error: 0.0449 380/380 [==============================] - 0s 74us/step - loss: 0.0216 - mean_squared_error: 0.0216 - val_loss: 0.0106 - val_mean_squared_error: 0.0106 Epoch 25/30 32/380 [=>............................] - ETA: 0s - loss: 0.0108 - mean_squared_error: 0.0108 380/380 [==============================] - 0s 73us/step - loss: 0.0226 - mean_squared_error: 0.0226 - val_loss: 0.0220 - val_mean_squared_error: 0.0220 Epoch 26/30 32/380 [=>............................] - ETA: 0s - loss: 0.0365 - mean_squared_error: 0.0365 380/380 [==============================] - 0s 73us/step - loss: 0.1305 - mean_squared_error: 0.1305 - val_loss: 0.0583 - val_mean_squared_error: 0.0583 Epoch 27/30 32/380 [=>............................] - ETA: 0s - loss: 0.0542 - mean_squared_error: 0.0542 380/380 [==============================] - 0s 75us/step - loss: 0.0318 - mean_squared_error: 0.0318 - val_loss: 0.0642 - val_mean_squared_error: 0.0642 Epoch 28/30 32/380 [=>............................] - ETA: 0s - loss: 0.0565 - mean_squared_error: 0.0565 380/380 [==============================] - 0s 73us/step - loss: 0.0287 - mean_squared_error: 0.0287 - val_loss: 0.0080 - val_mean_squared_error: 0.0080 Epoch 29/30 32/380 [=>............................] - ETA: 0s - loss: 0.0071 - mean_squared_error: 0.0071 380/380 [==============================] - 0s 74us/step - loss: 0.0274 - mean_squared_error: 0.0274 - val_loss: 0.0063 - val_mean_squared_error: 0.0063 Epoch 30/30 32/380 [=>............................] - ETA: 0s - loss: 0.0044 - mean_squared_error: 0.0044 380/380 [==============================] - 0s 75us/step - loss: 0.0063 - mean_squared_error: 0.0063 - val_loss: 0.0047 - val_mean_squared_error: 0.0047 Testing: Case True Estimate 0: 2.8966 2.9203 1: 3.5714 3.6538 2: 1.3504 1.3834 3: 4.4512 4.4424 4: 3.5437 3.6228 5: 2.2862 2.3170 6: 1.8434 1.8702 7: 1.4530 1.4644 8: 1.7830 1.7988 9: 3.6567 3.7465 10: 4.0334 4.1103 11: 2.9695 2.9830 12: 3.4010 3.4561 13: 2.1963 2.2375 14: 3.5508 3.6308 15: 0.9818 1.0824 16: 2.4629 2.4895 17: 2.0502 2.0993 18: 2.7961 2.8288 19: 4.4257 4.4239 20: 3.6132 3.6998 21: 4.8695 4.6974 22: 4.8942 4.7066 23: 4.9986 4.7375 24: 1.0783 1.1650 Graphics saved as "paraheat_gaussian_parameter.png" paraheat_gaussian_parameter: Normal end of execution. Tue May 7 10:30:06 2019