From cad015d0c1041e8e09bc6045a7cefae4ce4d4908 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?K=C3=BChle=2C=20Laura=20Christine=20=28lakue103=29?=
<laura.kuehle@uni-duesseldorf.de>
Date: Fri, 20 Aug 2021 17:14:15 +0200
Subject: [PATCH] Added ANN testing (based on Soraya's implementation).
---
ANN_Training.py | 66 ++++++++++++++++++++++++++++++++++++++++++++++---
Plotting.py | 24 ++++++++++++++++++
2 files changed, 87 insertions(+), 3 deletions(-)
diff --git a/ANN_Training.py b/ANN_Training.py
index 80ac4a0..4fab8f5 100644
--- a/ANN_Training.py
+++ b/ANN_Training.py
@@ -3,20 +3,23 @@
@author: Laura C. Kühle, Soraya Terrab (sorayaterrab)
TODO: Improve 'epoch_training()'
-TODO: Add ANN testing from Soraya
+TODO: Add ANN testing from Soraya -> Done
TODO: Add ANN classification from Soraya
TODO: Improve naming of training data/model (maybe different folders?)
TODO: Adjust input file naming to fit training data -> Done
TODO: Change code to add model directory if not existing -> Done
TODO: Remove unnecessary comments -> Done
+TODO: Add option to set plot directory
"""
import numpy as np
import os
import torch
from torch.utils.data import TensorDataset, DataLoader
+# from sklearn.metrics import accuracy_score, precision_recall_fscore_support
import ANN_Model
+from Plotting import plot_classification_accuracy
class ModelTrainer(object):
@@ -104,6 +107,62 @@ class ModelTrainer(object):
if valid_loss / len(valid_dl) < self._threshold:
break
+ def test_model(self):
+ self.epoch_training()
+ self._model.eval()
+
+ x_test, y_test = self._training_data['test']
+ model_output = torch.round(self._model(x_test.float()))
+ # acc = np.sum(model_output.numpy() == y_test.numpy())
+ test_accuracy = (model_output == y_test).float().mean()
+ print(test_accuracy)
+ # print(model_output.nelement())
+ # accuracy1 = torch.sum(torch.eq(model_output, y_test)).item() # /model_output.nelement()
+ # print(test_accuracy, accuracy1/model_output.nelement())
+ # print(accuracy1)
+
+ tp, fp, tn, fn = self._evaluate_classification(model_output, y_test)
+ precision, recall, accuracy = self._evaluate_stats(tp, fp, tn, fn)
+ # print(precision, recall)
+ # print(accuracy)
+ plot_classification_accuracy(precision, recall, accuracy, ['real'])
+
+ @staticmethod
+ def _evaluate_classification(model_output, true_output):
+ # Positive being Discontinuous/Troubled Cells, Negative being Smooth/Good Cells
+ true_positive = 0
+ true_negative = 0
+ false_positive = 0
+ false_negative = 0
+ for i in range(true_output.size()[0]):
+ if model_output[i, 1] == model_output[i, 0]:
+ print(i, model_output[i])
+ if true_output[i, 0] == torch.tensor([1]):
+ if model_output[i, 0] == true_output[i, 0]:
+ true_positive += 1
+ else:
+ false_negative += 1
+ if true_output[i, 1] == torch.tensor([1]):
+ if model_output[i, 1] == true_output[i, 1]:
+ true_negative += 1
+ else:
+ false_positive += 1
+
+ return true_positive, true_negative, false_positive, false_negative
+
+ @staticmethod
+ def _evaluate_stats(true_positive, true_negative, false_positive, false_negative):
+ if true_positive+false_positive == 0:
+ precision = 0
+ recall = 0
+ else:
+ precision = true_positive / (true_positive+false_positive)
+ recall = true_positive / (true_positive+false_negative)
+
+ accuracy = (true_positive+true_negative) / (true_positive+true_negative+false_positive+false_negative)
+ # print(true_positive+true_negative+false_positive+false_negative)
+ return precision, recall, accuracy
+
def save_model(self):
# Saving Model
train_name = self._training_file.split('.npy')[0]
@@ -124,6 +183,7 @@ class ModelTrainer(object):
# Loss Functions: BCELoss, BCEWithLogitsLoss, CrossEntropyLoss (not working), MSELoss (with reduction='sum')
# Optimizer: Adam, SGD
-trainer = ModelTrainer({'loss_function': 'MSELoss', 'loss_config': {'reduction': 'sum'}})
-trainer.epoch_training()
+trainer = ModelTrainer({'num_epochs': 100})
+# trainer.epoch_training()
+trainer.test_model()
trainer.save_model()
diff --git a/Plotting.py b/Plotting.py
index 5267856..2fa5fdb 100644
--- a/Plotting.py
+++ b/Plotting.py
@@ -2,6 +2,8 @@
"""
@author: Laura C. Kühle
+TODO: Give option to select plotting color
+
"""
import numpy as np
import matplotlib.pyplot as plt
@@ -112,3 +114,25 @@ def calculate_exact_solution(mesh, cell_len, wave_speed, final_time, interval_le
grid = np.reshape(np.array(grid), (1, len(grid) * len(grid[0])))
return grid, exact
+
+
+def plot_classification_accuracy(precision, recall, accuracy, xlabels):
+ precision = [precision]
+ recall = [recall]
+ accuracy = [accuracy]
+ pos = np.arange(len(xlabels))
+ width = 0.3
+ fig = plt.figure('classification_accuracy')
+ ax = fig.add_axes([0.15, 0.1, 0.75, 0.8])
+ ax.bar(pos - width, precision, width, label='Precision')
+ ax.bar(pos, recall, width, label='Recall')
+ ax.bar(pos + width, accuracy, width, label='Accuracy')
+ ax.set_xticks(x)
+ ax.set_xticklabels(xlabels)
+ ax.set_ylabel('Classification (%)')
+ ax.set_ylim(bottom=0.6)
+ ax.set_ylim(top=1.02)
+ ax.set_title('Non-Normalized Test Data')
+ ax.legend(loc='upper right')
+ # fig.tight_layout()
+ fig.savefig('TestAdamPrecisionRecallAccuracy.pdf')
--
GitLab