ML-Powered Security Log Monitoring for a Bank

Uncovering hidden security risks with machine learning-powered log monitoring for a bank

MADIS built a platform that collects and analyses logs from core banking systems, applications and infrastructure, and uses machine learning to detect anomalies.

Summary

MADIS built a security log monitoring platform for a bank that collects logs from core banking systems, applications and infrastructure, evaluates them against predefined security incident rules and applies machine learning to detect anomalies. The solution revealed significant security risks, markedly improved incident detection and response, and is still actively used.

The client

The client is a bank.

The challenge

The bank needed to collect and analyse logs from its core banking systems, applications and infrastructure, and evaluate them against predefined security incident rules. Security teams also needed a clear way to investigate: moving from a detected incident down to the underlying log entries.

Our approach

Log collection and rule-based evaluation

Logs from core banking systems, applications and infrastructure are collected and evaluated against predefined security incident rules.

Machine learning on the collected logs

Machine learning algorithms analyse the collected logs to detect anomalies.

Interactive dashboard

The processed data is visualised on an interactive dashboard, where users can drill down seamlessly from incidents to detailed logs.

The solution

The delivered platform combines the Elastic stack for collecting, processing and visualising logs with Python-based machine learning for anomaly detection.

Technologies: Elasticsearch · Logstash · Kibana · Python

Results

• Hidden risks found: machine learning detected anomalies that revealed significant security risks.
• Better detection and response: the rate of security incident detection and response improved markedly.
• Lasting value: the solution remains actively used today.

Facing a similar challenge?

Want to get more out of your security logs? Talk to our team about machine learning-based monitoring.