Upload complex_backend_security_with_encrypted_waves_and_anomaly_detection.py
Browse files
complex_backend_security_with_encrypted_waves_and_anomaly_detection.py
ADDED
@@ -0,0 +1,95 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# -*- coding: utf-8 -*-
|
2 |
+
"""Complex backend security with encrypted waves and anomaly detection
|
3 |
+
|
4 |
+
Automatically generated by Colab.
|
5 |
+
|
6 |
+
Original file is located at
|
7 |
+
https://colab.research.google.com/drive/1iCPhdZ1rZxxWmNqFPIUGGrtgVUfgcPcI
|
8 |
+
"""
|
9 |
+
|
10 |
+
from cryptography.fernet import Fernet
|
11 |
+
|
12 |
+
# Generate a key for encryption
|
13 |
+
key = Fernet.generate_key()
|
14 |
+
cipher = Fernet(key)
|
15 |
+
|
16 |
+
# Example wave data (as a list of floats)
|
17 |
+
wave_data = [0.1, 0.5, 0.3, 0.4, 0.9]
|
18 |
+
|
19 |
+
# Convert wave data to bytes and encrypt
|
20 |
+
wave_data_bytes = bytes(str(wave_data), 'utf-8')
|
21 |
+
encrypted_wave_data = cipher.encrypt(wave_data_bytes)
|
22 |
+
|
23 |
+
# Decrypting wave data
|
24 |
+
decrypted_wave_data_bytes = cipher.decrypt(encrypted_wave_data)
|
25 |
+
decrypted_wave_data = eval(decrypted_wave_data_bytes.decode('utf-8'))
|
26 |
+
|
27 |
+
print("Original Wave Data:", wave_data)
|
28 |
+
print("Encrypted Wave Data:", encrypted_wave_data)
|
29 |
+
print("Decrypted Wave Data:", decrypted_wave_data)
|
30 |
+
|
31 |
+
import torch
|
32 |
+
import torch.nn as nn
|
33 |
+
import torch.optim as optim
|
34 |
+
import numpy as np
|
35 |
+
|
36 |
+
class Autoencoder(nn.Module):
|
37 |
+
def __init__(self):
|
38 |
+
super(Autoencoder, self).__init__()
|
39 |
+
self.encoder = nn.Sequential(
|
40 |
+
nn.Linear(5, 3),
|
41 |
+
nn.ReLU(),
|
42 |
+
nn.Linear(3, 2),
|
43 |
+
nn.ReLU()
|
44 |
+
)
|
45 |
+
self.decoder = nn.Sequential(
|
46 |
+
nn.Linear(2, 3),
|
47 |
+
nn.ReLU(),
|
48 |
+
nn.Linear(3, 5),
|
49 |
+
nn.Sigmoid()
|
50 |
+
)
|
51 |
+
|
52 |
+
def forward(self, x):
|
53 |
+
x = self.encoder(x)
|
54 |
+
x = self.decoder(x)
|
55 |
+
return x
|
56 |
+
|
57 |
+
# Initialize the model, loss function, and optimizer
|
58 |
+
model = Autoencoder()
|
59 |
+
criterion = nn.MSELoss()
|
60 |
+
optimizer = optim.Adam(model.parameters(), lr=0.01)
|
61 |
+
|
62 |
+
normal_wave_data = torch.tensor([
|
63 |
+
[0.1, 0.2, 0.3, 0.4, 0.5],
|
64 |
+
[0.2, 0.3, 0.4, 0.5, 0.6],
|
65 |
+
[0.3, 0.4, 0.5, 0.6, 0.7]
|
66 |
+
], dtype=torch.float32)
|
67 |
+
|
68 |
+
# Training the model
|
69 |
+
for epoch in range(1000): # Example training loop
|
70 |
+
optimizer.zero_grad()
|
71 |
+
outputs = model(normal_wave_data)
|
72 |
+
loss = criterion(outputs, normal_wave_data)
|
73 |
+
loss.backward()
|
74 |
+
optimizer.step()
|
75 |
+
|
76 |
+
if (epoch+1) % 100 == 0:
|
77 |
+
print(f'Epoch [{epoch+1}/1000], Loss: {loss.item():.4f}')
|
78 |
+
|
79 |
+
# New wave data to check for anomalies
|
80 |
+
new_wave_data = torch.tensor([0.9, 0.8, 0.7, 0.6, 0.5], dtype=torch.float32)
|
81 |
+
|
82 |
+
# Reshape for single input
|
83 |
+
new_wave_data = new_wave_data.unsqueeze(0)
|
84 |
+
|
85 |
+
# Pass through the model
|
86 |
+
reconstructed_data = model(new_wave_data)
|
87 |
+
loss = criterion(reconstructed_data, new_wave_data)
|
88 |
+
|
89 |
+
# Set a threshold for anomaly detection
|
90 |
+
anomaly_threshold = 0.01
|
91 |
+
|
92 |
+
if loss.item() > anomaly_threshold:
|
93 |
+
print("Anomaly detected!")
|
94 |
+
else:
|
95 |
+
print("Data is normal.")
|