
Brain Connectivity Matrix Dataset
Brain-connectivity matrices for super-resolution research — recovering high-resolution connectomes from coarse scans. Released for anyone to build on.
// datasets · open science
Research is only real when it's reproducible and reusable. Here are the datasets I've released to the community — led by the Brain Connectivity Matrix Dataset on Kaggle, a CC0 public-domain contribution to super-resolution research in neuroscience.
// the flagship

Brain-connectivity matrices for super-resolution research — recovering high-resolution connectomes from coarse scans. Released for anyone to build on.
// the why
This dataset started with a simple question: can we recover a precise brain-wiring graph from a coarse, low-resolution scan? Super-resolution for connectomics. The matrices in this release are the raw material for that line of research — dense neural-connectivity matrices that let others train and test models predicting high-resolution connectomes from sparse observations.
It's also the seed of my connectomics-to-swarms thesis — the idea that I study intelligence from molecules to neurons to multi-agent systems. Releasing the data openly was the natural first step: a thesis is only credible when the evidence behind it is public.
// the how
Grab the files from the Kaggle page or the GitHub repo — no account gate, no form, no attribution required. The dataset is licensed CC0 (public domain): you can copy, modify, distribute, and build commercial products on it without asking permission or citing me.
If you do find it useful, a link back is appreciated (not required). The main ask is simply: use it to push connectomics and brain super-resolution forward.
// the bigger picture
This dataset sits at the center of the brain super-resolution project on my projects page. Same research thread, same source code — the release here is the data half, the models live there. Together they're the connective tissue between biology, machine learning, and the multi-agent systems I work on today.
Anonymized and aggregate datasets drawn from my agent-work evaluations. Honest note: these are not yet public — data privacy and anonymization for real agent workloads needs to be done carefully before I release anything. When they're ready, they'll live here.