noshot 1.0.0__py3-none-any.whl → 3.0.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (57) hide show
  1. noshot/data/ML TS XAI/ML/Main/1. EDA-PCA (Balance Scale Dataset).ipynb +139 -0
  2. noshot/data/ML TS XAI/ML/Main/1. EDA-PCA (Rice Dataset).ipynb +181 -0
  3. noshot/data/ML TS XAI/ML/Main/10. HMM Veterbi.ipynb +228 -0
  4. noshot/data/ML TS XAI/ML/Main/2. KNN (Balance Scale Dataset).ipynb +117 -0
  5. noshot/data/ML TS XAI/ML/Main/2. KNN (Iris Dataset).ipynb +165 -0
  6. noshot/data/ML TS XAI/ML/Main/2. KNN (Sobar-72 Dataset).ipynb +251 -0
  7. noshot/data/ML TS XAI/ML/Main/3. LDA (Balance Scale Dataset).ipynb +78 -0
  8. noshot/data/ML TS XAI/ML/Main/3. LDA (NPHA Doctor Visits Dataset).ipynb +114 -0
  9. noshot/data/ML TS XAI/ML/Main/4. Linear Regression (Machine Dataset).ipynb +115 -0
  10. noshot/data/ML TS XAI/ML/Main/4. Linear Regression (Real Estate Dataset).ipynb +159 -0
  11. noshot/data/ML TS XAI/{XAI/XAI 2/Exp-3 (EDA-loan).ipynb → ML/Main/5. Logistic Regression (Magic04 Dataset).ipynb } +53 -74
  12. noshot/data/ML TS XAI/ML/Main/5. Logistic Regression (Wine Dataset).ipynb +112 -0
  13. noshot/data/ML TS XAI/ML/Main/6. Naive Bayes Classifier (Agaricus Lepiota Dataset).ipynb +153 -0
  14. noshot/data/ML TS XAI/ML/Main/6. Naive Bayes Classifier (Wine Dataset).ipynb +89 -0
  15. noshot/data/ML TS XAI/ML/Main/7. SVM (Rice Dataset).ipynb +208 -0
  16. noshot/data/ML TS XAI/ML/Main/8. FeedForward NN (Sobar72 Dataset).ipynb +260 -0
  17. noshot/data/ML TS XAI/ML/Main/9. CNN (Cifar10 Dataset).ipynb +238 -0
  18. noshot/data/ML TS XAI/ML/Main/data/agaricus-lepiota.data +8124 -0
  19. noshot/data/ML TS XAI/ML/Main/data/balance-scale.txt +625 -0
  20. noshot/data/ML TS XAI/ML/Main/data/doctor-visits.csv +715 -0
  21. noshot/data/ML TS XAI/ML/Main/data/iris.csv +151 -0
  22. noshot/data/ML TS XAI/ML/Main/data/machine-data.csv +210 -0
  23. noshot/data/ML TS XAI/ML/Main/data/magic04.data +19020 -0
  24. noshot/data/ML TS XAI/ML/Main/data/real-estate.xlsx +0 -0
  25. noshot/data/ML TS XAI/ML/Main/data/rice.arff +3826 -0
  26. noshot/data/ML TS XAI/ML/Main/data/sobar-72.csv +73 -0
  27. noshot/data/ML TS XAI/ML/Main/data/wine-dataset.csv +179 -0
  28. noshot/data/ML TS XAI/ML/Other Codes.ipynb +158 -0
  29. noshot/data/ML TS XAI/ML/Rolls Royce AllinOne.ipynb +691 -0
  30. {noshot-1.0.0.dist-info → noshot-3.0.0.dist-info}/METADATA +1 -1
  31. noshot-3.0.0.dist-info/RECORD +38 -0
  32. {noshot-1.0.0.dist-info → noshot-3.0.0.dist-info}/WHEEL +1 -1
  33. noshot/data/ML TS XAI/TS/10. Seasonal ARIMA Forecasting.ipynb +0 -246
  34. noshot/data/ML TS XAI/TS/11. Multivariate ARIMA Forecasting.ipynb +0 -228
  35. noshot/data/ML TS XAI/TS/6. ACF PACF.ipynb +0 -77
  36. noshot/data/ML TS XAI/TS/7. Differencing.ipynb +0 -167
  37. noshot/data/ML TS XAI/TS/8. ARMA Forecasting.ipynb +0 -197
  38. noshot/data/ML TS XAI/TS/9. ARIMA Forecasting.ipynb +0 -220
  39. noshot/data/ML TS XAI/XAI/XAI 1/EDA2_chipsdatset.ipynb +0 -633
  40. noshot/data/ML TS XAI/XAI/XAI 1/EDA_IRISH_8thjan.ipynb +0 -326
  41. noshot/data/ML TS XAI/XAI/XAI 1/XAI_EX1 MODEL BIAS (FINAL).ipynb +0 -487
  42. noshot/data/ML TS XAI/XAI/XAI 1/complete_guide_to_eda_on_text_data.ipynb +0 -845
  43. noshot/data/ML TS XAI/XAI/XAI 1/deepchecksframeworks.ipynb +0 -100
  44. noshot/data/ML TS XAI/XAI/XAI 1/deepexplainers (mnist).ipynb +0 -90
  45. noshot/data/ML TS XAI/XAI/XAI 1/guidedbackpropagation.ipynb +0 -203
  46. noshot/data/ML TS XAI/XAI/XAI 1/updated_image_EDA1_with_LRP.ipynb +0 -3998
  47. noshot/data/ML TS XAI/XAI/XAI 1/zebrastripes.ipynb +0 -271
  48. noshot/data/ML TS XAI/XAI/XAI 2/EXP_5.ipynb +0 -1545
  49. noshot/data/ML TS XAI/XAI/XAI 2/Exp-3 (EDA-movie).ipynb +0 -229
  50. noshot/data/ML TS XAI/XAI/XAI 2/Exp-4(Flower dataset).ipynb +0 -237
  51. noshot/data/ML TS XAI/XAI/XAI 2/Exp-4.ipynb +0 -241
  52. noshot/data/ML TS XAI/XAI/XAI 2/Exp_2.ipynb +0 -352
  53. noshot/data/ML TS XAI/XAI/XAI 2/Exp_7.ipynb +0 -110
  54. noshot/data/ML TS XAI/XAI/XAI 2/FeatureImportance_SensitivityAnalysis.ipynb +0 -708
  55. noshot-1.0.0.dist-info/RECORD +0 -32
  56. {noshot-1.0.0.dist-info → noshot-3.0.0.dist-info}/licenses/LICENSE.txt +0 -0
  57. {noshot-1.0.0.dist-info → noshot-3.0.0.dist-info}/top_level.txt +0 -0
noshot/data/ML TS XAI/XAI/XAI 2/Exp-3 (EDA-movie).ipynb DELETED
@@ -1,229 +0,0 @@
1
- {
2
- "cells": [
3
- {
4
- "cell_type": "code",
5
- "execution_count": null,
6
- "id": "ccf7df96",
7
- "metadata": {},
8
- "outputs": [],
9
- "source": [
10
- "import pandas as pd \n",
11
- "import numpy as np\n",
12
- "import matplotlib.pyplot as plt\n",
13
- "import seaborn as sns"
14
- ]
15
- },
16
- {
17
- "cell_type": "code",
18
- "execution_count": null,
19
- "id": "05fcafeb",
20
- "metadata": {},
21
- "outputs": [],
22
- "source": [
23
- "auto=pd.read_csv(\"IMDB-Movie-Data.csv\")\n",
24
- "#auto.reset_index(inplace=True)\n",
25
- "auto.head()"
26
- ]
27
- },
28
- {
29
- "cell_type": "code",
30
- "execution_count": null,
31
- "id": "774d8ecf",
32
- "metadata": {},
33
- "outputs": [],
34
- "source": [
35
- "auto.head()"
36
- ]
37
- },
38
- {
39
- "cell_type": "code",
40
- "execution_count": null,
41
- "id": "66bef117",
42
- "metadata": {},
43
- "outputs": [],
44
- "source": [
45
- "auto.tail()"
46
- ]
47
- },
48
- {
49
- "cell_type": "code",
50
- "execution_count": null,
51
- "id": "d12303ac",
52
- "metadata": {},
53
- "outputs": [],
54
- "source": [
55
- "auto.info()"
56
- ]
57
- },
58
- {
59
- "cell_type": "code",
60
- "execution_count": null,
61
- "id": "13ca9a52",
62
- "metadata": {},
63
- "outputs": [],
64
- "source": [
65
- "auto.nunique()"
66
- ]
67
- },
68
- {
69
- "cell_type": "code",
70
- "execution_count": null,
71
- "id": "406b67f4",
72
- "metadata": {
73
- "scrolled": true
74
- },
75
- "outputs": [],
76
- "source": [
77
- "auto.isnull()"
78
- ]
79
- },
80
- {
81
- "cell_type": "code",
82
- "execution_count": null,
83
- "id": "eaa5f031",
84
- "metadata": {},
85
- "outputs": [],
86
- "source": [
87
- "auto.describe()"
88
- ]
89
- },
90
- {
91
- "cell_type": "code",
92
- "execution_count": null,
93
- "id": "214537c9",
94
- "metadata": {},
95
- "outputs": [],
96
- "source": [
97
- "data = auto.drop(['index'], axis = 1)\n",
98
- "data"
99
- ]
100
- },
101
- {
102
- "cell_type": "code",
103
- "execution_count": null,
104
- "id": "9e22f6d6",
105
- "metadata": {
106
- "scrolled": true
107
- },
108
- "outputs": [],
109
- "source": [
110
- "print(auto.Title.unique())"
111
- ]
112
- },
113
- {
114
- "cell_type": "code",
115
- "execution_count": null,
116
- "id": "95a9706b",
117
- "metadata": {},
118
- "outputs": [],
119
- "source": [
120
- "print(auto.Title.nunique())"
121
- ]
122
- },
123
- {
124
- "cell_type": "code",
125
- "execution_count": null,
126
- "id": "0c895e16",
127
- "metadata": {
128
- "scrolled": true
129
- },
130
- "outputs": [],
131
- "source": [
132
- "au=auto.head(50)\n",
133
- "au"
134
- ]
135
- },
136
- {
137
- "cell_type": "code",
138
- "execution_count": null,
139
- "id": "ff300c7a",
140
- "metadata": {
141
- "scrolled": true
142
- },
143
- "outputs": [],
144
- "source": [
145
- "plt.title(\"MOVIES REVENUES\")\n",
146
- "plt.xlabel(\"Title\")\n",
147
- "plt.ylabel(\"Revenue (Millions)\")\n",
148
- "plt.plot(au['Title'],au['Revenue (Millions)'],color=\"blue\")\n",
149
- "plt.show(10,20)"
150
- ]
151
- },
152
- {
153
- "cell_type": "code",
154
- "execution_count": null,
155
- "id": "f53a7f1d",
156
- "metadata": {},
157
- "outputs": [],
158
- "source": [
159
- "plt.title(\"MOVIES REVENUES\")\n",
160
- "plt.xlabel(\"Title\")\n",
161
- "plt.ylabel(\"Revenue (Millions)\")\n",
162
- "plt.bar(au['Title'],au['Revenue (Millions)'],color=\"green\")\n",
163
- "plt.show()\n"
164
- ]
165
- },
166
- {
167
- "cell_type": "code",
168
- "execution_count": null,
169
- "id": "f0625d78",
170
- "metadata": {},
171
- "outputs": [],
172
- "source": [
173
- "sns.pairplot(au)"
174
- ]
175
- },
176
- {
177
- "cell_type": "code",
178
- "execution_count": null,
179
- "id": "88d11bbd",
180
- "metadata": {},
181
- "outputs": [],
182
- "source": [
183
- "plt.hist(au['Revenue (Millions)'],bins=10,color='cyan')\n",
184
- "plt.show()"
185
- ]
186
- },
187
- {
188
- "cell_type": "code",
189
- "execution_count": null,
190
- "id": "bfc6edb5",
191
- "metadata": {},
192
- "outputs": [],
193
- "source": [
194
- "plt.figure(figsize=(12, 7))\n",
195
- "sns.heatmap(au.corr(), annot = True, vmin = -1, vmax = 1)\n",
196
- "plt.show()"
197
- ]
198
- },
199
- {
200
- "cell_type": "code",
201
- "execution_count": null,
202
- "id": "adc44fdc-323a-4b9f-9ec2-1227ef925241",
203
- "metadata": {},
204
- "outputs": [],
205
- "source": []
206
- }
207
- ],
208
- "metadata": {
209
- "kernelspec": {
210
- "display_name": "Python 3 (ipykernel)",
211
- "language": "python",
212
- "name": "python3"
213
- },
214
- "language_info": {
215
- "codemirror_mode": {
216
- "name": "ipython",
217
- "version": 3
218
- },
219
- "file_extension": ".py",
220
- "mimetype": "text/x-python",
221
- "name": "python",
222
- "nbconvert_exporter": "python",
223
- "pygments_lexer": "ipython3",
224
- "version": "3.12.4"
225
- }
226
- },
227
- "nbformat": 4,
228
- "nbformat_minor": 5
229
- }
noshot/data/ML TS XAI/XAI/XAI 2/Exp-4(Flower dataset).ipynb DELETED
@@ -1,237 +0,0 @@
1
- {
2
- "cells": [
3
- {
4
- "cell_type": "code",
5
- "execution_count": null,
6
- "id": "9c6bb3ac-e1f5-4b1e-8944-9fd6d95c5cf9",
7
- "metadata": {},
8
- "outputs": [],
9
- "source": [
10
- "import torch \n",
11
- "import torch.nn as nn \n",
12
- "import torch.optim as optim\n",
13
- "import torch.nn.functional as F\n",
14
- "import torchvision \n",
15
- "import torchvision.transforms as transforms\n",
16
- "import numpy as np\n",
17
- "import matplotlib.pyplot as plt \n",
18
- "from sklearn.tree import DecisionTreeClassifier\n",
19
- "from sklearn.tree import plot_tree"
20
- ]
21
- },
22
- {
23
- "cell_type": "code",
24
- "execution_count": null,
25
- "id": "11e04672-e757-459c-99d8-b100bf4e8a2a",
26
- "metadata": {},
27
- "outputs": [],
28
- "source": [
29
- "transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,), (0.5,))])\n",
30
- "trainset = torchvision.datasets.MNIST(root='126156009/17flowerdataset.zip', train=True, download=True, transform=transform)\n",
31
- "trainloader = torch.utils.data.DataLoader(trainset, batch_size=64, shuffle=True)\n",
32
- "\n",
33
- "testset = torchvision.datasets.MNIST(root='126156009/17flowerdataset.zip', train=False, download=True, transform=transform)\n",
34
- "testloader = torch.utils.data.DataLoader(testset, batch_size=64, shuffle=False)"
35
- ]
36
- },
37
- {
38
- "cell_type": "code",
39
- "execution_count": null,
40
- "id": "7107c76c-5ed7-43c9-a61d-4908cd06c062",
41
- "metadata": {},
42
- "outputs": [],
43
- "source": [
44
- "class CNN(nn.Module):\n",
45
- " def __init__(self):\n",
46
- " super(CNN, self).__init__()\n",
47
- " self.conv1 = nn.Conv2d(1, 16, kernel_size=3, padding=1)\n",
48
- " self.conv2 = nn.Conv2d(16, 32, kernel_size=3, padding=1)\n",
49
- " self.fc1 = nn.Linear(32 * 7 * 7, 128)\n",
50
- " self.fc2 = nn.Linear(128, 10)\n",
51
- "\n",
52
- " def forward(self, x):\n",
53
- " x = F.relu(self.conv1(x))\n",
54
- " x = F.max_pool2d(x, 2, 2)\n",
55
- " x = F.relu(self.conv2(x))\n",
56
- " x = F.max_pool2d(x, 2, 2)\n",
57
- " x = x.view(-1, 32 * 7 * 7)\n",
58
- " x = F.relu(self.fc1(x))\n",
59
- " x = self.fc2(x)\n",
60
- " return x\n",
61
- "\n",
62
- "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
63
- "model = CNN().to(device)\n",
64
- "criterion = nn.CrossEntropyLoss()\n",
65
- "optimizer = optim.Adam(model.parameters(), lr=0.001)"
66
- ]
67
- },
68
- {
69
- "cell_type": "code",
70
- "execution_count": null,
71
- "id": "9085fd94-535d-4e3d-9937-f3c5049a426d",
72
- "metadata": {},
73
- "outputs": [],
74
- "source": [
75
- "def train(model, trainloader, criterion, optimizer, epochs=3):\n",
76
- " model.train()\n",
77
- " for epoch in range(epochs):\n",
78
- " running_loss = 0.0\n",
79
- " for images, labels in trainloader:\n",
80
- " images, labels = images.to(device), labels.to(device)\n",
81
- " optimizer.zero_grad()\n",
82
- " outputs = model(images) \n",
83
- " \n",
84
- " print(f\"Logits for first image in batch: {outputs[0]}\") \n",
85
- "\n",
86
- " loss = criterion(outputs, labels)\n",
87
- " loss.backward()\n",
88
- " optimizer.step()\n",
89
- " running_loss += loss.item()\n",
90
- " print(f\"Epoch {epoch + 1}, Loss: {running_loss / len(trainloader):.4f}\")\n",
91
- "\n",
92
- "train(model, trainloader, criterion, optimizer)\n"
93
- ]
94
- },
95
- {
96
- "cell_type": "code",
97
- "execution_count": null,
98
- "id": "27e125ef-9b2d-4768-b097-ff905f3cbcc8",
99
- "metadata": {},
100
- "outputs": [],
101
- "source": [
102
- "def extract_features(model, dataloader):\n",
103
- " model.eval()\n",
104
- " features, labels = [], []\n",
105
- " with torch.no_grad():\n",
106
- " for images, lbls in dataloader:\n",
107
- " images = images.to(device)\n",
108
- " outputs = model(images) \n",
109
- "\n",
110
- " print(f\"Logits for first image in batch: {outputs[0]}\") \n",
111
- "\n",
112
- " features.extend(outputs.cpu().numpy()) \n",
113
- " labels.extend(lbls.numpy())\n",
114
- " return np.array(features), np.array(labels)\n",
115
- "\n",
116
- "X_train, y_train = extract_features(model, trainloader)\n",
117
- "X_test, y_test = extract_features(model, testloader)"
118
- ]
119
- },
120
- {
121
- "cell_type": "code",
122
- "execution_count": null,
123
- "id": "6807872d-6d75-40fb-a168-21b110c0f168",
124
- "metadata": {},
125
- "outputs": [],
126
- "source": [
127
- "dt = DecisionTreeClassifier(max_depth=5) \n",
128
- "dt.fit(X_train, y_train)\n",
129
- "\n",
130
- "\n",
131
- "acc = dt.score(X_test, y_test)\n",
132
- "print(f\"Surrogate Model Accuracy: {acc * 100:.2f}%\")\n"
133
- ]
134
- },
135
- {
136
- "cell_type": "code",
137
- "execution_count": null,
138
- "id": "d57a082e-4d32-48ed-a0bc-d8bb1e6b2e08",
139
- "metadata": {},
140
- "outputs": [],
141
- "source": [
142
- "\n",
143
- "def visualize_surrogate_model(dt):\n",
144
- " plt.figure(figsize=(20, 16))\n",
145
- " plot_tree(dt, filled=True, feature_names=[f\"Feature {i}\" for i in range(X_train.shape[1])], class_names=[str(i) for i in range(10)], rounded=True,fontsize=14)\n",
146
- " plt.title(\"Surrogate Model - Decision Tree\")\n",
147
- " plt.show()\n",
148
- "\n",
149
- "visualize_surrogate_model(dt)\n",
150
- "\n",
151
- "\n",
152
- "def plot_feature_importance(dt, feature_names):\n",
153
- " feature_importances = dt.feature_importances_\n",
154
- " indices = np.argsort(feature_importances)[::-1]\n",
155
- "\n",
156
- " plt.figure(figsize=(10, 6))\n",
157
- " plt.title(\"Feature Importances (Surrogate Model)\")\n",
158
- " plt.barh(range(X_train.shape[1]), feature_importances[indices], align=\"center\")\n",
159
- " plt.yticks(range(X_train.shape[1]), [f\"Feature {i}\" for i in indices])\n",
160
- " plt.xlabel(\"Importance\")\n",
161
- " plt.show()\n",
162
- "\n",
163
- "plot_feature_importance(dt, [f\"Feature {i}\" for i in range(X_train.shape[1])])\n",
164
- "\n",
165
- "\n",
166
- "def visualize_feature_maps(model, input_image):\n",
167
- " model.eval()\n",
168
- " layers = [model.conv1, model.conv2]\n",
169
- " activations = []\n",
170
- "\n",
171
- " def save_activation(name):\n",
172
- " def hook(model, input, output):\n",
173
- " activations.append(output)\n",
174
- " return hook\n",
175
- "\n",
176
- " \n",
177
- " hooks = []\n",
178
- " for layer in layers:\n",
179
- " hooks.append(layer.register_forward_hook(save_activation(layer.__class__.__name__)))\n",
180
- "\n",
181
- " \n",
182
- " input_image = input_image.unsqueeze(0).to(device)\n",
183
- " model(input_image)\n",
184
- "\n",
185
- " \n",
186
- " for i, activation in enumerate(activations):\n",
187
- " activation = activation.squeeze(0).cpu().detach().numpy()\n",
188
- " num_filters = activation.shape[0]\n",
189
- "\n",
190
- " \n",
191
- " fig, axes = plt.subplots(1, num_filters, figsize=(15, 8))\n",
192
- " for j in range(num_filters):\n",
193
- " axes[j].imshow(activation[j], cmap='gray')\n",
194
- " axes[j].axis('off')\n",
195
- " axes[j].set_title(f\"Filter {j + 1}\")\n",
196
- " plt.show()\n",
197
- "\n",
198
- " \n",
199
- " for hook in hooks:\n",
200
- " hook.remove()\n",
201
- "\n",
202
- "\n",
203
- "sample_image, sample_label = testset[0]\n",
204
- "visualize_feature_maps(model, sample_image)"
205
- ]
206
- },
207
- {
208
- "cell_type": "code",
209
- "execution_count": null,
210
- "id": "f6f63b48-11b5-4c46-adf8-9c504d6256c8",
211
- "metadata": {},
212
- "outputs": [],
213
- "source": []
214
- }
215
- ],
216
- "metadata": {
217
- "kernelspec": {
218
- "display_name": "Python 3 (ipykernel)",
219
- "language": "python",
220
- "name": "python3"
221
- },
222
- "language_info": {
223
- "codemirror_mode": {
224
- "name": "ipython",
225
- "version": 3
226
- },
227
- "file_extension": ".py",
228
- "mimetype": "text/x-python",
229
- "name": "python",
230
- "nbconvert_exporter": "python",
231
- "pygments_lexer": "ipython3",
232
- "version": "3.12.4"
233
- }
234
- },
235
- "nbformat": 4,
236
- "nbformat_minor": 5
237
- }
noshot/data/ML TS XAI/XAI/XAI 2/Exp-4.ipynb DELETED
@@ -1,241 +0,0 @@
1
- {
2
- "cells": [
3
- {
4
- "cell_type": "code",
5
- "execution_count": null,
6
- "id": "a956d239-b4b0-4e65-ac1a-d8047cfc883f",
7
- "metadata": {},
8
- "outputs": [],
9
- "source": [
10
- "import torch \n",
11
- "import torch.nn as nn \n",
12
- "import torch.optim as optim\n",
13
- "import torch.nn.functional as F\n",
14
- "import torchvision \n",
15
- "import torchvision.transforms as transforms\n",
16
- "import numpy as np\n",
17
- "import matplotlib.pyplot as plt \n",
18
- "from sklearn.tree import DecisionTreeClassifier\n",
19
- "from sklearn.tree import plot_tree"
20
- ]
21
- },
22
- {
23
- "cell_type": "code",
24
- "execution_count": null,
25
- "id": "7a0d4f98-bfb5-40a0-9c0e-8843361cb7a7",
26
- "metadata": {},
27
- "outputs": [],
28
- "source": [
29
- "transform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,), (0.5,))])\n",
30
- "trainset = torchvision.datasets.MNIST(root='./data', train=True, download=True, transform=transform)\n",
31
- "trainloader = torch.utils.data.DataLoader(trainset, batch_size=64, shuffle=True)\n",
32
- "\n",
33
- "testset = torchvision.datasets.MNIST(root='./data', train=False, download=True, transform=transform)\n",
34
- "testloader = torch.utils.data.DataLoader(testset, batch_size=64, shuffle=False)"
35
- ]
36
- },
37
- {
38
- "cell_type": "code",
39
- "execution_count": null,
40
- "id": "7c323e6e-58bf-498a-bc60-1e02a1aea4ef",
41
- "metadata": {},
42
- "outputs": [],
43
- "source": [
44
- "class CNN(nn.Module):\n",
45
- " def __init__(self):\n",
46
- " super(CNN, self).__init__()\n",
47
- " self.conv1 = nn.Conv2d(1, 16, kernel_size=3, padding=1)\n",
48
- " self.conv2 = nn.Conv2d(16, 32, kernel_size=3, padding=1)\n",
49
- " self.fc1 = nn.Linear(32 * 7 * 7, 128)\n",
50
- " self.fc2 = nn.Linear(128, 10)\n",
51
- "\n",
52
- " def forward(self, x):\n",
53
- " x = F.relu(self.conv1(x))\n",
54
- " x = F.max_pool2d(x, 2, 2)\n",
55
- " x = F.relu(self.conv2(x))\n",
56
- " x = F.max_pool2d(x, 2, 2)\n",
57
- " x = x.view(-1, 32 * 7 * 7)\n",
58
- " x = F.relu(self.fc1(x))\n",
59
- " x = self.fc2(x)\n",
60
- " return x\n",
61
- "\n",
62
- "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")\n",
63
- "model = CNN().to(device)\n",
64
- "criterion = nn.CrossEntropyLoss()\n",
65
- "optimizer = optim.Adam(model.parameters(), lr=0.001)"
66
- ]
67
- },
68
- {
69
- "cell_type": "code",
70
- "execution_count": null,
71
- "id": "57e73bc1-cbfb-46ca-b5d2-4798b8b59a89",
72
- "metadata": {},
73
- "outputs": [],
74
- "source": [
75
- "def train(model, trainloader, criterion, optimizer, epochs=3):\n",
76
- " model.train()\n",
77
- " for epoch in range(epochs):\n",
78
- " running_loss = 0.0\n",
79
- " for images, labels in trainloader:\n",
80
- " images, labels = images.to(device), labels.to(device)\n",
81
- " optimizer.zero_grad()\n",
82
- " outputs = model(images) # CNN outputs (logits)\n",
83
- "\n",
84
- " # Print CNN output (logits) for the first image in the batch\n",
85
- " print(f\"Logits for first image in batch: {outputs[0]}\") # First image in the batch\n",
86
- "\n",
87
- " loss = criterion(outputs, labels)\n",
88
- " loss.backward()\n",
89
- " optimizer.step()\n",
90
- " running_loss += loss.item()\n",
91
- " print(f\"Epoch {epoch + 1}, Loss: {running_loss / len(trainloader):.4f}\")\n",
92
- "\n",
93
- "train(model, trainloader, criterion, optimizer)\n"
94
- ]
95
- },
96
- {
97
- "cell_type": "code",
98
- "execution_count": null,
99
- "id": "93b6189f-889e-419e-95d6-9154c2acca26",
100
- "metadata": {},
101
- "outputs": [],
102
- "source": [
103
- "def extract_features(model, dataloader):\n",
104
- " model.eval()\n",
105
- " features, labels = [], []\n",
106
- " with torch.no_grad():\n",
107
- " for images, lbls in dataloader:\n",
108
- " images = images.to(device)\n",
109
- " outputs = model(images) # CNN outputs (logits)\n",
110
- "\n",
111
- " # Print CNN outputs (logits) for the first image in each batch\n",
112
- " print(f\"Logits for first image in batch: {outputs[0]}\") # This will print the logits for the first image\n",
113
- "\n",
114
- " features.extend(outputs.cpu().numpy()) # Extract CNN outputs as features\n",
115
- " labels.extend(lbls.numpy())\n",
116
- " return np.array(features), np.array(labels)\n",
117
- "\n",
118
- "X_train, y_train = extract_features(model, trainloader)\n",
119
- "X_test, y_test = extract_features(model, testloader)"
120
- ]
121
- },
122
- {
123
- "cell_type": "code",
124
- "execution_count": null,
125
- "id": "44442993-376b-41c0-9dc9-f55cfdc0268c",
126
- "metadata": {},
127
- "outputs": [],
128
- "source": [
129
- "dt = DecisionTreeClassifier(max_depth=5) \n",
130
- "dt.fit(X_train, y_train)\n",
131
- "\n",
132
- "\n",
133
- "acc = dt.score(X_test, y_test)\n",
134
- "print(f\"Surrogate Model Accuracy: {acc * 100:.2f}%\")\n"
135
- ]
136
- },
137
- {
138
- "cell_type": "code",
139
- "execution_count": null,
140
- "id": "79bb7a9f-b4ba-4bc1-95ef-ecd1cf525b45",
141
- "metadata": {},
142
- "outputs": [],
143
- "source": [
144
- "\n",
145
- "def visualize_surrogate_model(dt):\n",
146
- " plt.figure(figsize=(12, 8))\n",
147
- " plot_tree(dt, filled=True, feature_names=[f\"Feature {i}\" for i in range(X_train.shape[1])], class_names=[str(i) for i in range(10)], rounded=True)\n",
148
- " plt.title(\"Surrogate Model - Decision Tree\")\n",
149
- " plt.show()\n",
150
- "\n",
151
- "visualize_surrogate_model(dt)\n",
152
- "\n",
153
- "\n",
154
- "def plot_feature_importance(dt, feature_names):\n",
155
- " feature_importances = dt.feature_importances_\n",
156
- " indices = np.argsort(feature_importances)[::-1]\n",
157
- "\n",
158
- " plt.figure(figsize=(10, 6))\n",
159
- " plt.title(\"Feature Importances (Surrogate Model)\")\n",
160
- " plt.barh(range(X_train.shape[1]), feature_importances[indices], align=\"center\")\n",
161
- " plt.yticks(range(X_train.shape[1]), [f\"Feature {i}\" for i in indices])\n",
162
- " plt.xlabel(\"Importance\")\n",
163
- " plt.show()\n",
164
- "\n",
165
- "\n",
166
- "plot_feature_importance(dt, [f\"Feature {i}\" for i in range(X_train.shape[1])])\n",
167
- "\n",
168
- "\n",
169
- "\n",
170
- "def visualize_feature_maps(model, input_image):\n",
171
- " model.eval()\n",
172
- " layers = [model.conv1, model.conv2]\n",
173
- " activations = []\n",
174
- "\n",
175
- " def save_activation(name):\n",
176
- " def hook(model, input, output):\n",
177
- " activations.append(output)\n",
178
- " return hook\n",
179
- "\n",
180
- " \n",
181
- " hooks = []\n",
182
- " for layer in layers:\n",
183
- " hooks.append(layer.register_forward_hook(save_activation(layer.__class__.__name__)))\n",
184
- "\n",
185
- " \n",
186
- " input_image = input_image.unsqueeze(0).to(device)\n",
187
- " model(input_image)\n",
188
- "\n",
189
- " \n",
190
- " for i, activation in enumerate(activations):\n",
191
- " activation = activation.squeeze(0).cpu().detach().numpy()\n",
192
- " num_filters = activation.shape[0]\n",
193
- "\n",
194
- " \n",
195
- " fig, axes = plt.subplots(1, num_filters, figsize=(15, 8))\n",
196
- " for j in range(num_filters):\n",
197
- " axes[j].imshow(activation[j], cmap='gray')\n",
198
- " axes[j].axis('off')\n",
199
- " axes[j].set_title(f\"Filter {j + 1}\")\n",
200
- " plt.show()\n",
201
- "\n",
202
- " \n",
203
- " for hook in hooks:\n",
204
- " hook.remove()\n",
205
- "\n",
206
- "\n",
207
- "sample_image, sample_label = testset[0]\n",
208
- "visualize_feature_maps(model, sample_image)"
209
- ]
210
- },
211
- {
212
- "cell_type": "code",
213
- "execution_count": null,
214
- "id": "d572de18-1df8-41b8-8b24-618790d7d0aa",
215
- "metadata": {},
216
- "outputs": [],
217
- "source": []
218
- }
219
- ],
220
- "metadata": {
221
- "kernelspec": {
222
- "display_name": "Python 3 (ipykernel)",
223
- "language": "python",
224
- "name": "python3"
225
- },
226
- "language_info": {
227
- "codemirror_mode": {
228
- "name": "ipython",
229
- "version": 3
230
- },
231
- "file_extension": ".py",
232
- "mimetype": "text/x-python",
233
- "name": "python",
234
- "nbconvert_exporter": "python",
235
- "pygments_lexer": "ipython3",
236
- "version": "3.12.4"
237
- }
238
- },
239
- "nbformat": 4,
240
- "nbformat_minor": 5
241
- }