Milton

Milton

@miltonllera.bsky.social

Former code monkey turned prompt monkey

64 followers108 following28 posts

Milton
@miltonllera.bsky.social

Our work on adapting NCAs towards generating controlled morphological changes is now in preprint!

Mike Levin@drmichaellevin.bsky.social

new #preprint: t.co/eE1qqGSflK Ben Hartl explains: In 1917, D'Arcy Thompson drew Cartesian grids around different anatomical shapes & showed that deforming the grid could change one species' shape into that of another. We find that "reusable regulatory handles control phenotypic variation" in NCAs

t.cohttps://arxiv.org/abs/2609.29755
View full thread
Milton
@miltonllera.bsky.social

6/n Most surprisingly, we find that the pre-patterns are not used to directly approximate the target. Indeed, even though more complex SIREN functions support better target generation, the initial patterns are not closer to the target when we increase complexity

Milton
@miltonllera.bsky.social

7/n Thus we show that the system can offer a valuable insight related to the functional utility of pre-patterns. Furthermore we can use it to replicate how embryogenesis combines growth and self-organisation. Check out the preprint here: arxiv.org/abs/2605.14998

arxiv.orgLearning Developmental Scaffoldings to Guide Self-OrganisationFrom subcellular structures to entire organisms, many natural systems generate complex organisation through self-organisation: local interactions that collectively give rise to global structure withou...
View full thread
Milton
@miltonllera.bsky.social

4/n To test the first result we trained the model using different levels of reliability in the cell update and compared to a standard NCA which can only self-organise. The results show that as we decrease cell reliability, pre-patterns help with robustness:

Milton
@miltonllera.bsky.social

5/n Using a goal-directed NCA formulation, we show that for similar parameter counts, using pre-patterns leads to systems that can encode more patterns:

View full thread
Milton
@miltonllera.bsky.social

2/n To study this question we paired an NCA with a learned spatial pre-pattern generator in the form of an implicit representation model, specifically a SIREN. The pre-pattern acts as a scaffolding which sets initial conditions of the NCA but cannot reproduce the target directly

Milton
@miltonllera.bsky.social

3/n Our results show 3 things: 1. Pre-patterns improve robustness, 2. they increase model capacity, 3. and, perhaps surprisingly, are not used to directly approximate the target (or some approximation of it)

Milton
@miltonllera.bsky.social

1/n Self-organisation is studied as a mechanism that works in isolation, unfolding development from random patterns. But biology rarely works this way; embryos start with spatial information guiding development. In this work we model this computationally arxiv.org/abs/2605.14998

Milton reposted

ALICE Workshop
@aliceworkshop.org

What a week—projects, tutorials, discussion groups, art demo & even a complexity-alife quiz! Thank you to all the attendees who brought a fantastic energy to this year's ALICE workshop. We can't wait to see you again next year, this time in Norway 🇳🇴 Thank you to @aicentre.dk and @itu.dk !

Milton
@miltonllera.bsky.social

Excited about self-organizing systems and their likely synergies with modern approaches to AI? Then submit to the SONI special session at #ALife2025

Elias Najarro@najarro.science

Announcing SONI: Self-Organising Neural Intelligence(s) 🧠 An #ALife2025 special session to bring together AI & ALife communities to explore how self-organisation can contribute to neural paradigms of intelligence. Call for papers sites.google.com/view/soni-al... |Submission deadline: May 4th

sites.google.comSelf-Organising Neural Intelligence(s)Important dates: Submission Deadline: May 4th, 2025 Author Notification: TBA

You are up to date.