Create app.py

app.py

@@ -0,0 +1,138 @@
+import streamlit as st
+import torch
+import torch.nn as nn
+import numpy as np
+import joblib
+import plotly.graph_objects as go
+
+# -----------------------------
+# Page config (premium look)
+# -----------------------------
+st.set_page_config(
+    page_title="Cyber Threat Detection Dashboard",
+    page_icon="🛡️",
+    layout="wide"
+)
+
+# -----------------------------
+# Title Section
+# -----------------------------
+st.markdown("""
+# 🛡️ Cyber Threat Detection Dashboard
+**Deep Learning–Based Suspicious Activity Detection**
+""")
+
+st.markdown("---")
+
+# -----------------------------
+# Model definition
+# -----------------------------
+class SuspiciousModel(nn.Module):
+    def __init__(self, input_dim):
+        super().__init__()
+        self.net = nn.Sequential(
+            nn.Linear(input_dim, 128),
+            nn.BatchNorm1d(128),
+            nn.ReLU(),
+            nn.Dropout(0.3),
+
+            nn.Linear(128, 64),
+            nn.BatchNorm1d(64),
+            nn.ReLU(),
+            nn.Dropout(0.3),
+
+            nn.Linear(64, 1)
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+# -----------------------------
+# Load model & scaler
+# -----------------------------
+scaler = joblib.load("scaler.pkl")
+INPUT_DIM = scaler.mean_.shape[0]
+
+model = SuspiciousModel(INPUT_DIM)
+model.load_state_dict(torch.load("model.pth", map_location="cpu"))
+model.eval()
+
+# -----------------------------
+# Sidebar – Log Input
+# -----------------------------
+st.sidebar.header("🔍 Log Feature Input")
+
+features = []
+for i in range(INPUT_DIM):
+    val = st.sidebar.number_input(
+        f"Feature {i+1}",
+        value=0.0,
+        step=0.01
+    )
+    features.append(val)
+
+# -----------------------------
+# Prediction
+# -----------------------------
+if st.sidebar.button("🚨 Analyze Event"):
+    x = np.array(features).reshape(1, -1)
+    x_scaled = scaler.transform(x)
+    x_tensor = torch.tensor(x_scaled, dtype=torch.float32)
+
+    with torch.no_grad():
+        prob = torch.sigmoid(model(x_tensor)).item()
+
+    # -------------------------
+    # Risk level logic
+    # -------------------------
+    if prob > 0.7:
+        risk = "HIGH"
+        color = "red"
+        action = "Immediate investigation required. Isolate affected system."
+    elif prob > 0.4:
+        risk = "MEDIUM"
+        color = "orange"
+        action = "Monitor closely. Correlate with other logs."
+    else:
+        risk = "LOW"
+        color = "green"
+        action = "No action required. Log for auditing."
+
+    # -------------------------
+    # Main dashboard layout
+    # -------------------------
+    col1, col2, col3 = st.columns(3)
+
+    col1.metric("Risk Level", risk)
+    col2.metric("Suspicion Probability", f"{prob:.2f}")
+    col3.metric("Recommended Action", action)
+
+    # -------------------------
+    # Gauge chart (VERY impressive)
+    # -------------------------
+    fig = go.Figure(go.Indicator(
+        mode="gauge+number",
+        value=prob * 100,
+        title={'text': "Threat Confidence (%)"},
+        gauge={
+            'axis': {'range': [0, 100]},
+            'bar': {'color': color},
+            'steps': [
+                {'range': [0, 40], 'color': "lightgreen"},
+                {'range': [40, 70], 'color': "orange"},
+                {'range': [70, 100], 'color': "red"}
+            ],
+        }
+    ))
+
+    st.plotly_chart(fig, use_container_width=True)
+
+    # -------------------------
+    # Explanation block
+    # -------------------------
+    st.markdown("### 🧠 Model Interpretation")
+    st.info(
+        "The model analyzes behavioral patterns in system logs and assigns a risk score "
+        "based on learned representations of malicious activity from the BETH dataset."
+    )
+