Most AI pipelines throw away structure and meaning to compress data.
I built something that doesn’t.

What I Built: A Lossless, Structure-Preserving Matrix Intelligence Engine

Use it to:

  • Find connections between datasets (e.g., drugs ↔ genes ↔ categories)
  • Analyze matrix structure (sparsity, binary, diagonal)
  • Cluster semantically similar datasets
  • Benchmark reconstruction (up to 100% accuracy)

No AI guessing — just explainable structure-preserving math.

Key Benchmarks (Real Biomedical Data)

Try It Instantly (Docker Only)

Just run this — no setup required:

bashCopyEditmkdir data results
# Drop your TSV/CSV files into the data folder
docker run -it \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/results:/app/results \
  fikayomiayodele/hyperdimensional-connection

Your results show up in the results/folder.

Installation, Usage & Documentation

All installation instructions and usage examples are in the GitHub README:
📘 github.com/fikayoAy/MatrixTransformer

No Python dependencies needed — just Docker.
Runs on Linux, macOS, Windows, or GitHub Codespaces for browser-only users.

📄 Scientific Paper

This project is based on the research papers:

Ayodele, F. (2025). Hyperdimensional connection method – A Lossless Framework Preserving Meaning, Structure, and Semantic Relationships across Modalities.(A MatrixTransformer subsidiary). Zenodo. https://doi.org/10.5281/zenodo.16051260

Ayodele, F. (2025). MatrixTransformer. Zenodo. https://doi.org/10.5281/zenodo.15928158

It includes full benchmarks, architecture, theory, and reproducibility claims.

🧬 Use Cases

  • Drug Discovery: Build knowledge graphs from drug–gene–category data
  • ML Pipelines: Select algorithms based on matrix structure
  • ETL QA: Flag isolated or corrupted files instantly
  • Semantic Clustering: Without any training
  • Bio/NLP/Vision Data: Works on anything matrix-like

💡 Why This Is Different

Feature Traditional Tools This Tool
Deep learning required ✅ ❌ (deterministic math)
Semantic relationships ❌ ✅ 99.999%+ similarity
Cross-domain support ❌ ✅ (bio, text, visual)
100% reproducible ❌ ✅ (same results every time)
Zero setup ❌ ✅ Docker-only

🤝 Join In or Build On It

If you find it useful:

  • 🌟 Star the repo
  • 🔁 Fork or extend it
  • 📎 Cite the paper in your own work
  • 💬 Drop feedback or ideas—I’m exploring time-series & vision next

This is open source, open science, and meant to empower others.

📦 Docker Hub: fikayomiayodele/hyperdimensional-connection
🧠 GitHub: github.com/fikayoAy/MatrixTransformer

Looking forward to feedback from researchers, skeptics, and builders

Posted by Hyper_graph

3 comments

  1. Idk if it’s just me reading in bed too late but nothing in the “paper” or this post makes any sense at all lol

    I think you gotta lay off the chatgpt

  2. I’m a little confused by the papers you’ve submitted. This isn’t my field at all, but I *do* have a number of years of experience in academia so I’ve ended up with some questions:

    * The Zenodo papers haven’t been peer-reviewed. Have you submitted them to any peer-reviewed journals?
    * Is the raw data from either paper available? I could not see it.
    * Why have you referenced no other authors than yourself? This is unheard-of in genuine research.
    * You’ve clearly put some effort into the appearance of your paper, but your references section doesn’t use a consistent referencing convention. Why is this?

    It’d take an expert in the field to really assess the software itself, but these appear to be red flags on a general level.

Comments are closed.