THREATBYTESTM research
Research outputs
Outputs include publications, datasets, software releases, and thesis chapters, documenting contribution to predictive malware intelligence and autonomous defence knowledge.
Contribution to knowledge01
The programme advances:
- Agentic, cognitive, reasoning-driven models that learn proactively from historical, simulated, and community-submitted malware.
- A predictive intelligence framework inferring intent and enabling pre-execution mitigation.
- Quantitative evidence comparing reasoning-enabled predictive SI to reactive detection for zero-day, polymorphic, and adaptive threats.
- An open, extensible architecture suitable for enterprise and national defence integration with continuous learning from emerging ecosystems.
- Academic understanding of safe reinforcement learning and reasoning in real-time detection and autonomous mitigation.
- Collaborative intelligence showing how community submissions enhance adaptive learning.Findings target academic literature, policy frameworks, and next-generation systems that transform cybersecurity from reactive detection to autonomous, predictive defence.
Output types02
Public summaries may link to restricted full text, datasets, or model artefacts inside the authenticated workspace when access policy allows.