Sat Seeker is a research project developing anomaly detection methods for massive datasets, from satellite imagery to radio signals. Powered by stochastic physics and optimised search algorithms. Currently in active modelling phase.
The Problem
Whether scanning radio frequencies for anomalous signals or analysing satellite imagery for environmental changes, the fundamental challenge is the same: detecting rare, meaningful patterns in datasets too large for traditional analysis.
Current methods rely on either brute force grid search (slow and expensive) or machine learning (requires extensive training data and produces unexplainable results). Neither is well suited for detecting genuinely novel anomalies that don't match any known pattern.
We are exploring whether physics-based stochastic analysis can solve this more efficiently: searching smarter, not harder, with fully explainable results.
Our Approach
Sat Seeker uses the same foundational approach behind all Innova Castle projects: physics-based modelling and mathematical analysis applied to complex detection problems.
Explores large data spaces more efficiently than traditional grid search methods using stochastic optimisation
Specialised for anomalies that don't match known patterns, without requiring training data
Results grounded in stochastic analysis with clear physical reasoning, not black box scores
Works on new data immediately without historical training sets or labelled examples
Active Research Lines
The same core methodology applied to two distinct detection challenges.
Signal Classification · Technosignatures
Framework for classifying radio signals as natural or artificial using stochastic analysis. Distinguishes potential technosignatures from cosmic noise with physics-based criteria, no training data required.
Sentinel-2 · Climate · Deforestation
Satellite imagery analysis for environmental monitoring. Detects deforestation, drought patterns, and land use changes using Sentinel-2 data with optimised search algorithms.
Where We Are
Sat Seeker is in active modelling and testing phase across both research lines.
Whether you're exploring risk analytics for your organisation or interested in our research, we're always open to a conversation.
