Welcome to YTUnity

One free SpiderTaz Network account works across every network site.

Featured creators

Creators on the SpiderTaz Network

Become a creator →

AI Anomaly Detection Demo: Using Deep Learning to Find Network Threats

Infosec Institute Guide 1 months ago

Description

Watch Dwayne Roberts’ complete session, the hands-on AI cyber readiness workshop and all other AI Cyber Readiness Summit sessions on demand through September 25, 2026:
https://www.infosecinstitute.com/events/ai-cybersecurity-summit/?utm_source=youtube&utm_medium=video&utm_campaign=aisummit

Try the notebook from this demonstration:
https://colab.research.google.com/drive/1WxaELKluxX9ghy4oSBvkgKlzl4qMWs9R?usp=sharing

Signature-based detection is effective at identifying known threats. But what happens when network activity doesn’t match something your tools have seen before?

In this technical demo from the Infosec Institute AI Cyber Readiness Summit, Senior Technical Trainer Dwayne Roberts explores how deep learning and variational autoencoders (VAEs) can help identify anomalous network behavior.

Using a Google Colab notebook and synthetic network traffic, Dwayne demonstrates how a model can learn patterns of normal behavior, use reconstruction error to flag anomalies and balance detection sensitivity against false positives. He also shows how an anomaly detection model could supplement an existing security stack rather than replace signature-based detection.

In this demo, you’ll see:

0:00 — Why signature-based detection can miss unknown threats
1:01 — Walking through the anomaly detection notebook
1:46 — Network features used in the model
2:05 — Generating normal and attack traffic
3:10 — How the VAE reconstructs normal behavior
3:30 — Building and training the model
4:17 — Evaluating training loss
5:36 — Setting anomaly detection thresholds
6:21 — Visualizing reconstruction error
7:26 — Comparing normal and anomalous traffic
8:09 — Evaluating detection and false positives
8:48 — Reading the confusion matrix
10:33 — Visualizing normal vs. attack traffic
11:08 — How this could fit into a real-world detection stack

Watch Dwayne’s complete session and explore the full AI Cyber Readiness Summit while on-demand access remains available through September 25:
https://www.infosecinstitute.com/events/ai-cybersecurity-summit/?utm_source=youtube&utm_medium=video&utm_campaign=aisummit

About Infosec
Infosec Institute makes cyber workforce readiness a reality. Through role-based training that spans certification boot camps, skills development, security awareness and human risk management, we develop the capabilities your entire workforce needs to stay ahead of today's threats. The result: SOC analysts defending the perimeter, security engineers building secure systems and front-desk employees spotting advanced phishing attempts. 70% of the Fortune 500 have trusted Infosec for more than two decades. Learn more at infosecinstitute.com.