Spaces:
Sleeping
Sleeping
File size: 7,709 Bytes
8ee5513 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 | """Visualize federated learning results."""
import json
import os
def plot_accuracy_over_rounds(results: list, output_path: str = "federated_results.png"):
"""Plot accuracy, F1, precision, recall over federated rounds.
Shows:
- Line chart: accuracy, F1, precision, recall per round
- Bar chart: privacy budget consumed
"""
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(14, 6))
rounds = [r["round"] + 1 for r in results]
accuracies = [r["accuracy"] for r in results]
f1_scores = [r["f1"] for r in results]
precisions = [r["precision"] for r in results]
recalls = [r["recall"] for r in results]
# Plot 1: Metrics over rounds
ax1.plot(rounds, accuracies, "o-", color="#0D9488", linewidth=2,
markersize=6, label="Accuracy")
ax1.plot(rounds, f1_scores, "s-", color="#F59E0B", linewidth=2,
markersize=6, label="F1 Score")
ax1.plot(rounds, precisions, "^-", color="#3B82F6", linewidth=2,
markersize=5, label="Precision")
ax1.plot(rounds, recalls, "v-", color="#8B5CF6", linewidth=2,
markersize=5, label="Recall")
# Annotate first and last accuracy
ax1.annotate(
f"{accuracies[0]:.2f}",
xy=(rounds[0], accuracies[0]),
xytext=(rounds[0] + 0.3, accuracies[0] - 0.04),
fontsize=9, color="#0D9488",
)
ax1.annotate(
f"{accuracies[-1]:.2f}",
xy=(rounds[-1], accuracies[-1]),
xytext=(rounds[-1] - 0.8, accuracies[-1] + 0.03),
fontsize=9, color="#0D9488",
)
ax1.set_xlabel("Federated Round")
ax1.set_ylabel("Score")
ax1.set_title("MLP Model Improvement Over Federated Rounds")
ax1.legend(loc="lower right")
ax1.grid(True, alpha=0.3)
ax1.set_ylim(0.0, 1.05)
ax1.set_xticks(rounds)
# Plot 2: Privacy budget over rounds
epsilons = [r.get("epsilon_spent", 0) for r in results]
if any(e > 0 for e in epsilons):
bars = ax2.bar(rounds, epsilons, color="#EF4444", alpha=0.7)
ax2.set_ylabel("Cumulative Privacy Budget (\u03b5)")
for bar, eps in zip(bars, epsilons):
ax2.text(
bar.get_x() + bar.get_width() / 2,
bar.get_height() + 0.01,
f"{eps:.2f}",
ha="center", va="bottom", fontsize=8,
)
else:
ax2.text(0.5, 0.5, "No DP applied", ha="center", va="center",
transform=ax2.transAxes, fontsize=14, color="gray")
ax2.set_xlabel("Federated Round")
ax2.set_title("Privacy Budget Consumption")
ax2.grid(True, alpha=0.3)
ax2.set_xticks(rounds)
plt.tight_layout()
plt.savefig(output_path, dpi=150, bbox_inches="tight")
print(f"Saved plot to {output_path}")
plt.close()
def plot_dp_comparison(comparison: dict, output_path: str = "dp_comparison.png"):
"""Plot accuracy with DP vs without DP side by side.
Args:
comparison: dict with keys 'with_dp' and 'without_dp',
each a list of round result dicts.
output_path: Where to save the figure.
"""
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
results_dp = comparison["with_dp"]
results_no_dp = comparison["without_dp"]
rounds_dp = [r["round"] + 1 for r in results_dp]
rounds_no_dp = [r["round"] + 1 for r in results_no_dp]
fig, axes = plt.subplots(1, 2, figsize=(14, 6))
# Plot 1: Accuracy comparison
ax = axes[0]
ax.plot(rounds_dp, [r["accuracy"] for r in results_dp],
"o-", color="#EF4444", linewidth=2, markersize=6,
label="With DP (\u03b5=0.3)")
ax.plot(rounds_no_dp, [r["accuracy"] for r in results_no_dp],
"s-", color="#0D9488", linewidth=2, markersize=6,
label="Without DP")
ax.set_xlabel("Federated Round")
ax.set_ylabel("Accuracy")
ax.set_title("Accuracy: DP vs No-DP")
ax.legend()
ax.grid(True, alpha=0.3)
ax.set_ylim(0.0, 1.05)
ax.set_xticks(rounds_dp)
# Annotate final values
acc_dp_final = results_dp[-1]["accuracy"]
acc_no_dp_final = results_no_dp[-1]["accuracy"]
ax.annotate(f"{acc_dp_final:.3f}", xy=(rounds_dp[-1], acc_dp_final),
xytext=(rounds_dp[-1] - 1.5, acc_dp_final - 0.05),
fontsize=9, color="#EF4444")
ax.annotate(f"{acc_no_dp_final:.3f}", xy=(rounds_no_dp[-1], acc_no_dp_final),
xytext=(rounds_no_dp[-1] - 1.5, acc_no_dp_final + 0.03),
fontsize=9, color="#0D9488")
# Plot 2: F1 comparison
ax = axes[1]
ax.plot(rounds_dp, [r["f1"] for r in results_dp],
"o-", color="#EF4444", linewidth=2, markersize=6,
label="With DP (\u03b5=0.3)")
ax.plot(rounds_no_dp, [r["f1"] for r in results_no_dp],
"s-", color="#0D9488", linewidth=2, markersize=6,
label="Without DP")
ax.set_xlabel("Federated Round")
ax.set_ylabel("F1 Score")
ax.set_title("F1 Score: DP vs No-DP")
ax.legend()
ax.grid(True, alpha=0.3)
ax.set_ylim(0.0, 1.05)
ax.set_xticks(rounds_dp)
plt.tight_layout()
plt.savefig(output_path, dpi=150, bbox_inches="tight")
print(f"Saved DP comparison plot to {output_path}")
plt.close()
def print_summary(results: list):
"""Print a text summary of simulation results."""
print("\n" + "=" * 60)
print("FEDERATED LEARNING SIMULATION SUMMARY")
print("=" * 60)
for r in results:
print(f"\nRound {r['round']+1}:")
print(f" Accuracy: {r['accuracy']:.4f}")
print(f" F1 Score: {r['f1']:.4f}")
print(f" Precision: {r.get('precision', 0):.4f}")
print(f" Recall: {r.get('recall', 0):.4f}")
print(f" Devices: {r['n_devices']}")
eps = r.get('epsilon_spent', 0)
if eps > 0:
print(f" \u03b5 spent: {eps:.4f}")
first = results[0]
last = results[-1]
acc_delta = last["accuracy"] - first["accuracy"]
f1_delta = last["f1"] - first["f1"]
print(
f"\nImprovement: accuracy {first['accuracy']:.4f} -> "
f"{last['accuracy']:.4f} ({acc_delta:+.4f})"
)
print(
f" F1 {first['f1']:.4f} -> "
f"{last['f1']:.4f} ({f1_delta:+.4f})"
)
if __name__ == "__main__":
results_path = os.path.join(
os.path.dirname(os.path.abspath(__file__)), "simulation_results.json"
)
if os.path.exists(results_path):
with open(results_path) as f:
data = json.load(f)
# Handle both formats: list (single run) or dict (comparison)
if isinstance(data, list):
results = data
print_summary(results)
plot_accuracy_over_rounds(
results,
output_path=os.path.join(
os.path.dirname(os.path.abspath(__file__)),
"federated_results.png",
),
)
elif isinstance(data, dict):
# Comparison format
if "with_dp" in data:
print("\n--- WITH DP ---")
print_summary(data["with_dp"])
print("\n--- WITHOUT DP ---")
print_summary(data["without_dp"])
plot_dp_comparison(
data,
output_path=os.path.join(
os.path.dirname(os.path.abspath(__file__)),
"dp_comparison.png",
),
)
else:
print("Unknown results format.")
else:
print("No simulation results found. Run simulate.py first.")
|