<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>SYNAPSE // SEC</title>
    <link>https://devsecmed.com/</link>
    <description>Adversarial Robustness in Clinical Vision Models, Zero-Trust AI Infrastructure &amp; Systems Security.</description>
    <language>en-us</language>
    <atom:link href="https://devsecmed.com/rss.xml" rel="self" type="application/rss+xml" />
    
    <item>
      <title>Privacy-Preserving Federated Learning in Hospital Networks: Mitigating Gradient Inversion Attacks</title>
      <link>https://devsecmed.com/blog/privacy-preserving-federated-learning-gradient-inversion-defense</link>
      <guid isPermaLink="true">https://devsecmed.com/blog/privacy-preserving-federated-learning-gradient-inversion-defense</guid>
      <pubDate>Mon, 02 Nov 2026 09:15:00 GMT</pubDate>
      <description>Sharing raw parameter gradients in multi-hospital federated learning does not guarantee patient privacy: optimization-based gradient inversion can reconstruct diagnostic chest X-rays and volumetric slices. We derive tight Rényi Differential Privacy (RDP) guarantees with adaptive gradient clipping.</description>
      <category>ML Systems &amp; Privacy</category>
    </item>
    <item>
      <title>Memory-Safe Sandboxing &amp; eBPF Runtime Monitoring for Untrusted Clinical AI &amp; Agent Execution</title>
      <link>https://devsecmed.com/blog/memory-safe-sandboxing-ebpf-runtime-monitoring-ai-execution</link>
      <guid isPermaLink="true">https://devsecmed.com/blog/memory-safe-sandboxing-ebpf-runtime-monitoring-ai-execution</guid>
      <pubDate>Sat, 31 Oct 2026 10:30:00 GMT</pubDate>
      <description>Clinical AI pipelines and autonomous security agents routinely parse untrusted binary formats (DICOM, NIfTI, ONNX, safetensors) and invoke native C/C++ shared libraries. We design a zero-trust Linux sandbox combining Rust Landlock LSM with Linux Security Module (LSM) eBPF ring-buffer enforcement.</description>
      <category>Cybersecurity</category>
    </item>
    <item>
      <title>Adversarial Robustness in 3D Volumetric Medical Imaging: Defending Vision Transformers Against Imperceptible Perturbations</title>
      <link>https://devsecmed.com/blog/adversarial-robustness-3d-volumetric-medical-imaging-vit</link>
      <guid isPermaLink="true">https://devsecmed.com/blog/adversarial-robustness-3d-volumetric-medical-imaging-vit</guid>
      <pubDate>Wed, 28 Oct 2026 14:00:00 GMT</pubDate>
      <description>Volumetric 3D Swin-UNETR segmentation models achieve clinician-level Dice scores on multi-modal brain MRI, yet remain brittle under L-infinity perturbations below sensor thermal noise. We derive Lipschitz-constrained volumetric attention and evaluate certified smoothing bounds.</description>
      <category>Medical AI</category>
    </item>
  </channel>
</rss>