// AI news · 2026-11-15 · Nature
Explainable IoT Intrusion Detection via Bio-Inspired Encoding
A bio-inspired encoding framework transforms network flow features into amino acid sequences, summarized by ten protein structural properties.
Published 2026-11-15 08:00 (Beijing time)Source NatureAuto-summarised, fact-checked against source
01Summary
Tested on CICIoT2023, most models achieve 98.4–98.5% accuracy with low false alarm rates. The RNN model reaches 98.5% accuracy with fewer than 15,000 parameters and 0.1 ms per flow. Key decision attributes identified are β-Turn and Isoelectric Point, strongly correlated with packet count and size.
Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.
02Key points
- Flow features encoded into amino acid sequences via bio-inspired method.
- Models achieve 98.4–98.5% accuracy on CICIoT2023 dataset.
- RNN model has <15k parameters and 0.1ms per flow inference time.
- β-Turn and Isoelectric Point are key decision attributes linked to packet size and count.
03Source
Explainable IoT intrusion detection via bio-inspired encoding and interpretable AI models
Nature · published 2026-11-15 08:00 Beijing time
05Related news
- Same source · Dual Expert ML for Twin Tunnel Settlement Prediction2026-11-15 · Nature
- Same source · HGS-BioAbductor: Explainable Survival Modeling and Hypothesis Abduction2026-11-14 · Nature
- Same topic · Cohere Finds Self-Hosted Open-Weight Models Need Egress Controls2026-11-15 · InfoQ
- Same topic · Bowen Yang Returns as Jensen Huang on SNL2026-11-15 · Deadline
- Same day · Tech Execs Petition for Nvidia to Reinstatement of Jensen Huang2026-11-15 · TechCrunch AI
From AIMap AI news: collected from 1,100+ official and media sources. Summaries only restate the original article; follow the link for the full story. All stories of 2026-11-15 · Archive by date · All AI news · Today