At Howard Science Limited, we draw on decades of AI innovation rooted in defence and security applications. Founded in October 2008 by Dr Daniel Howard, a former Capability Leader in AI at the UK Defence Evaluation and Research Agency (DERA) and Company Fellow of QinetiQ, our work bridges deep theoretical advances with operational solutions for defence, security, and healthcare.
Core Competency Areas
1. Signal Processing & Data Recovery
We developed the proprietary Stochastic Bernstein functions, an advanced signal processing technique combining deconvolution and convolution with adaptive smoothing. Originally funded by the UK MOD, it enables:
- Noise filtering and denoising in complex environments
- Super-resolution imaging for security and surveillance
- Image compression alternatives to JPEG (relevant to steganography and steganalysis)
- Selective image zooming that respects hard/soft edges
Other applications include:
- Spectral data recovery (e.g. MALDI-TOF)
- Evolution of analytical solutions to non-self-adjoint differential equations.
2. Medical Imaging & AI in Mammography
In collaboration with world-leading radiologist Dr László Tabár, we pioneered:
- The first AI-based mammogram taxonomy system, classifying breast types based on tissue structure by means of an Unsupervised learning (ART-2 Neural Networks).
Applications and significance:
- Personalised screening strategies
- Clinical negligence evaluations
- Correlations between genomics, lifestyle factors, and cancer risk
3. Genetic Programming & Explainable AI
We co-developed Multirun Subtree Encapsulation (MSTE) with John Koza and Simon Roberts:
- Solved previously intractable AI problems like parity.
- Provides hierarchically structured, explainable AI solutions, ideal for defence and high-stakes domains.
This directly informs our NATO DIANA proposal for trustworthy AI decision support.
4. Traffic Prediction & Infrastructure Analytics
Partnered with the UK Highways Agency to deliver:
- AI models for motorway incident prediction and journey time estimation using MIDAS data.
- Explainable AI models, providing interpretable outputs for traffic engineers.
Outcomes contributed to active traffic management practices widely used today.
5. Genome Discovery
Applied evolutionary computation via evolved finite state machines to gene and regulatory sequence discovery in the human genome, contributing to bioinformatics innovation.
6. Defence Strategy Modelling & Human Factors
- Modelled drug trade dynamics and social networks for the former DSTL’s Counter Terrorism Theory Centre.
- Simulated soldier patrol effectiveness in Afghanistan.
- Published strategic insights on terrorism and collaborated on armour research.
7. Biomedical Signal Analysis
Developed AI tools for ECG analysis in pharmaceutical trials, supporting cardiac anomaly detection crucial for regulatory submissions like FDA reviews.
8. Computer Vision for Defence
Pioneered machine vision methods for target detection using Neural Networks and Genetic Programming—foundational work for current autonomous surveillance systems by the RAF and others.
Translating Legacy to Modern Solutions
These capabilities feed directly into our current projects:
- NATO DIANA Proposal 1: Human Resilience AI Co-Pilot — integrating neuroscience and cybernetic principles to enhance cognitive resilience under stress.
- NATO DIANA Proposal 2: Explainable Genetic Programming for Drone Manoeuvre Autonomy — evolving lightweight, certifiable pursuit-evasion strategies for contested airspace.
- DASA Proposal: Shadow Biometrics — drone-based gait recognition and identity inference from shadows.
We continue to translate deep theoretical expertise into practical, explainable, and secure AI for defence and national security.
Optional Access to Publications
For a curated selection of technical papers and legacy research, please contact us via our Contact Page or request access to our password-protected archive.
Howard Science Limited — Pioneering AI from theory to battlefield resilience.