A new way of doing sience

.Simple, intuitive and easy to learn
A method that can survive hundreds of years
without computers or infrastructure but also
.creates powerful, decentralised models

I have 5 minutes to learn



External articles

How to build Digital Hippocampus
Data Layering (data science)

Read More On

Build for everyone

Does not rely on understanding of maths or statistics

Learning Superstar

!Learn, uplearn, scale
Algorithm is incredibly dynamic and agile if you need it to be

Hive Mind

Use large models or build milions of interconnected and lean expert models


Energy efficient

You can scale your models
on the fly - using smaller in places like IoT or mobile devices


aboutimg
How it works

Become Machine Learning
Expert in just 5 minutes

Watch

I. Construction rules

- Imagine you have a stream of data in an Array
- Array is limited by maxLength
- You have multiple arrays, each one a bit shorter

II. Bahaviour Rules

- Each new element goes to the longest array
- With each new piece, remove the last bit
- Search for duplicates within an array
- Add duplicate to the next (shorter) array

III. Indexing Rules

- Data will form a pyramide
- Objects on the top are patterns (the most common)
- Longest array is just a short term memory
- Assign weights to bits accordingly


Easy to learn

powerful in practice

Where to use it - Case Studies

How to build projects - tutorials

Marie Maps
Advanced Patterns Recognition

aboutimg

Marie Maps is an xray for cities

Our PoC is collecting visual data from Open Street Map
translating them into patterns, scaterred across a city

This project heavily benefits from patterns recognition. Visual icons has been extracted from the map tiles and collected into an information hub

City maps tiled & crawled

After this treatment, each tile (256x256 pixels, representing roughly 10 square meters) has a separate model (as a weights map json file).

Project can search through similar tiles to highlight patterns or match one place with others on a map and project alternatives

Visit website

Demo Video



A-Irene
Cybersec & Anomaly Detection

aboutimg

Detecting on-line traffic disruptions

This project has been inspired by an alert fatigue, plaguing cybersec industry. We started small, by analysing websites and servers disruptions, instantly alerting owners

With A-Irene, marketers can save clients from burning their money on campaigns while landing pages crashed or loading really slow

Visit website

Demo Video



Marie Tiles
On-device Machine Learning

aboutimg

IoT-ready classifiers

AI puzzle pieces to build powerfull and smart applications

While Digital Hippocampus is a scalable concept, we've used
it to build a framework, constructed from micro classifiers
you can stack together to build smart applications

Visit website

Demo Video



powerful in practice

Easy to learn

How to build projects - tutorials
aboutimg
Tutorials comming soon
Marcin Rybicki

My Manifesto

Blackout-proof AI devices

?Why can't we teach advanced AI in an elementary school
?Who said advanced AI can't be simple
?Why not to build AI engines as mechanical tools

I believe we can set a new direction
for computer science

Why corporate powerhouses should dominate data science
Why predictions and forecasting should be a domain of scientists
Why your browser can't simply train a model on a device

mobile
user_img-lg
Explainable AI Architect

Marcin Rybicki

Hi, I'm an ex game developer and
a law school dropout
I made my first real money in advertising, designing brands, creating advert games

I was always fascinated by an algorithms and optimisations, creating new oportunities to deliver efficiency. When I was twelve years old I started conceptualising a formula to rule all computatons and AI

In 2017 - at age 37 - I took a year off from work to develop Marie AI - a distributed machine learning architecture, based on Digital Hippocampus

Since then, I'm on a quest to commercialise and prove to the world this technology works as good - if not better - than traditional neutral networks

FAQs

Frequently Asked Questions

In a human brain, Hippocampus governs the field of memory, defining which information going from short term to a long term. Basically - in my view - Hippocampus is a tool that curates informations, sorting out the noise. My creation - Digital Hippocampus - does exatly that. Once an array pyramice is formed, you have a noise on the bottom, patterns on the top
It's a first proven and tested solution that can be taught in elementary school as well as applied in the most advanced models. Does not require any knowledge of mathematics just intelligence and creativity, opening new oportunities
Digital Hippocampus simplifies the way we do dictionairy compression (ie. LZW), allowing for complex compression phrases to be used in cloud storage
It's not a commercial project. It's a mission aimed to educate people about new directions we can take when it comes to data science and machine learning
Smart people have their ways to make money (and it's not OF)
This project is non-profit. I believe we deserve alternatives to neural networks and humongous costs of running machine learning models

Follow Us and say Hi

LinkedIn
YouTube

whitepaper
Contact Us

Get in touch

  • Company (IP owner)

    Spruce Mobile LTD, London UK


    Digital Hippocampus was concieved
    and made in Ipswich, UK

  • Email

    Yourmail@gmail.com

  • Company's website

    spruce-mobile.com

  • Sister Project

    marieai.com