Re: Categorical Foundations of Generative AI [reader]

Thank you Milton
Yes, it can be generalized to address broader challenges
What I think is important here, is actually not category theory per se
(which has its limitation)
but that Gen AI is being anchored into KR/ontology of sorts
that is what this CG has been advoating all along and is nice to see happen

On Mon, Jun 24, 2024 at 6:08 PM Milton Ponson <rwiciamsd@gmail.com> wrote:

> Seems like more people have been exploring the same concepts. This GAIA
> model restricts itself to generative AI, but it can be generalized using
> category theory to all forms of knowledge representations.
>
> Combining constructibility theory, category theory, symplectic geometry
> and algebraic topology we can define universes of discourse that cover most
> of the knowledge representation methodologies using generalized graph
> concepts.
>
> Milton Ponson
> Rainbow Warriors Core Foundation
> CIAMSD Institute-ICT4D Program
> +2977459312
> PO Box 1154, Oranjestad
> Aruba, Dutch Caribbean
>
>
> On Fri, Jun 21, 2024 at 11:43 PM Paola Di Maio <paola.dimaio@gmail.com>
> wrote:
>
>>
>> Greetings,  W3C AI KR CG,  Happy Solstice
>>
>> a great way to start the summer by reading this paper
>> In my view this is a contribution towards neurosymbolic AI/KR and a clear
>> signal
>>
>> Enjoy
>> ----------------------------------------------------------------
>>
>> GAIA: Categorical Foundations of Generative AI
>> by Sridhar Mahadeva
>> https://arxiv.org/pdf/2402.18732
>>
>> In this paper, we propose GAIA, a generative AI architecture based on
>> category theory. GAIA is based on a hierarchical model where modules are
>> organized as a simplicial complex. Each simplicial complex updates its
>> internal parameters biased on information it receives from its superior
>> simplices and in turn relays updates to its subordinate sub-simplices.
>> Parameter updates are formulated in terms of lifting diagrams over
>> simplicial sets, where inner and outer horn extensions correspond to
>> different types of learning problems. Backpropagation is modeled as an
>> endofunctor over the category of parameters, leading to a coalgebraic
>> formulation of deep learning.
>>
>