Vuneravility Analyst

In this example, we first load data from the KDD Cup 1999 dataset, which contains labeled network activity logs. Then, we perform the necessary data preprocessing, such as removing non-relevant columns and encoding categorical variables.

After normalizing the data, we build an Autoencoder model using the TensorFlow Keras library. The Autoencoder is trained to reconstruct the input data with the least possible loss. Next, we calculate the reconstruction error on the training data and set a threshold for anomaly detection based on a high percentile of the reconstruction error.

Finally, we use the set threshold to predict anomalies in the test data and evaluate the performance of the model using classification metrics. This advanced approach allows us to detect intrusions in a network environment using unsupervised learning techniques based on Autoencoders.

vulnerabilidad5.py

@@ -0,0 +1,51 @@
+import pandas as pd
+import numpy as np
+from sklearn.preprocessing import StandardScaler
+from sklearn.model_selection import train_test_split
+from tensorflow.keras.models import Sequential
+from tensorflow.keras.layers import Dense
+
+# Cargar los datos del conjunto de datos KDD Cup 1999
+data = pd.read_csv("kddcup.csv")
+
+# Preprocesamiento de datos
+# Eliminar columnas no relevantes y codificar variables categóricas
+
+# Separar características y etiquetas
+X = data.drop("label", axis=1)
+y = data["label"]
+
+# Normalizar los datos
+scaler = StandardScaler()
+X_scaled = scaler.fit_transform(X)
+
+# Dividir los datos en conjuntos de entrenamiento y prueba
+X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=42)
+
+# Construir el modelo de Autoencoder
+model = Sequential([
+    Dense(64, activation='relu', input_shape=(X_train.shape[1],)),
+    Dense(32, activation='relu'),
+    Dense(64, activation='relu'),
+    Dense(X_train.shape[1], activation='sigmoid')
+])
+
+model.compile(optimizer='adam', loss='mse')
+
+# Entrenar el modelo de Autoencoder
+model.fit(X_train, X_train, epochs=10, batch_size=32, validation_data=(X_test, X_test))
+
+# Reconstruir los datos de entrada y calcular el error de reconstrucción
+X_train_pred = model.predict(X_train)
+train_mse = np.mean(np.power(X_train - X_train_pred, 2), axis=1)
+
+# Establecer un umbral para la detección de anomalías
+threshold = np.percentile(train_mse, 95)
+
+# Predecir anomalías en los datos de prueba
+X_test_pred = model.predict(X_test)
+test_mse = np.mean(np.power(X_test - X_test_pred, 2), axis=1)
+y_pred = (test_mse > threshold).astype(int)
+
+# Imprimir métricas de evaluación del modelo
+print(classification_report(y_test, y_pred))