mastodon.ie is one of the many independent Mastodon servers you can use to participate in the fediverse.
Irish Mastodon - run from Ireland, we welcome all who respect the community rules and members.

Administered by:

Server stats:

1.6K
active users

#hyperdimensionalcomputing

0 posts0 participants0 posts today

I get really sick of hearing about OpenAI and the rest of the LLM crowd making absurd claims about LLMs like ChatGPT being AGI. LLMs are not even close to AGI. If you want to see real research into AGI, watch this video on Chris Eliasmith's Spaun (semantic pointer architecture unified network). It may not look as snazzy and consumer-ready as all these hallucinating LLMs being marketed to us, but this is the real deal: A single system that can do lots of different things the brain does and that LLMs can never do.

youtu.be/I5h-xjddzlY?si=a-dS3y

#Spaun
#ArtificialGeneralIntelligence #AGI #AI #ML
#SemanticPointerArchitecture #SPA
#Nengo
#Hypervectors
#VectorSymbolicArchitecture #VSA
#HyperdimensionalComputing #HDC
#HolographicReducedRepresentation #HRR
#OpenAI #ChatGPT #LLM #LLMs

Replied in thread

@root The closest technologies we have to how the human brain works are not LLMs, but some less well-known ones: reinforcement learning algorithms and hyperdimensional computing. If you want to see what HDC is capable of, check out this video:

youtu.be/P_WRCyNQ9KY?si=JgAuOJ

#HDC #HyperdimensionalComputing
#VSA #VectorSymbolicArchitecture
#HRR #HolographicReducedRepresentation
#SpikingNeuralNetworks
#AGI #ArtificialGeneralIntelligence
#LLM

The next VSAonline webinar is at 17:00 UTC (not the usual time), Monday 27 January.

Zoom: ltu-se.zoom.us/j/65564790287

WEB: bit.ly/vsaonline

Speaker: Anthony Thomas from UC Davis, USA

Title: ”Sketching a Picture of Vector Symbolic Architectures”

Abstract : Sketching algorithms are a broad area of research in theoretical computer science and numerical analysis that aim to distil data into a simple summary, called a "sketch," that retains some essential notion of structure while being much more efficient to store, query, and transmit.

Vector-symbolic architectures (VSAs) are an approach to computing on data represented using random vectors, and provide an elegant conceptual framework for realizing a wide variety of data structures and algorithms in a way that lends itself to implementation in highly-parallel and energy-efficient computer hardware.

Sketching algorithms and VSA have a substantial degree of consonance in their methods, motivations, and applications. In this tutorial style talk, I will discuss some of the connections between these two fields, focusing, in particular, on the connections between VSA and tensor-sketches, a family of sketching algorithms concerned with the setting in which the data being sketched can be decomposed into Kronecker (tensor) products between more primitive objects. This is exactly the situation of interest in VSA and the two fields have arrived at strikingly similar solutions to this problem.

#VectorSymbolicArchitectures #VSA #HyperdimensionalComputing #HDC #AI #ML #ComputationalCognitiveScience #CompCogSci #MathematicalPsychology #MathPsych #CognitiveScience #CogSci @cogsci

Zoom VideoJoin our Cloud HD Video MeetingZoom is the leader in modern enterprise video communications, with an easy, reliable cloud platform for video and audio conferencing, chat, and webinars across mobile, desktop, and room systems. Zoom Rooms is the original software-based conference room solution used around the world in board, conference, huddle, and training rooms, as well as executive offices and classrooms. Founded in 2011, Zoom helps businesses and organizations bring their teams together in a frictionless environment to get more done. Zoom is a publicly traded company headquartered in San Jose, CA.

I was looking through the VSAonline website (sites.google.com/view/hdvsaonl) and saw that by the end of this year there will be 96 recorded webinars plus the recordings from the Midnight Sun 2023 workshop.

So, if you're interested in #VectorSymbolicArchitecture / #HyperdimensionalComputing or just #VSA / #HDC -curious, the VSAonline website is the place to go (or head over to hd-computing.com/home, where there's a wider range of resources, including publication lists).

sites.google.comVSAONLINEIf you want to give a credit to this webinar series use the following entry when citing (BibTeX). Vector Symbolic Architectures (VSA), which is also known as hyperdimensional computing (HDC) is a subclass of cognitive computing architectures, which is rooted in the observation that key aspects of

Here's a really interesting (long) paper on what a theory of computing based on arbitrary physical substrates might look like: arxiv.org/abs/2307.15408

"Toward a formal theory for computing machines made out of whatever physics offers: extended version"

Herbert Jaeger, Beatriz Noheda, Wilfred G. van der Wiel (2023)

@bnoheda

arXiv.orgToward a formal theory for computing machines made out of whatever physics offers: extended versionApproaching limitations of digital computing technologies have spurred research in neuromorphic and other unconventional approaches to computing. Here we argue that if we want to systematically engineer computing systems that are based on unconventional physical effects, we need guidance from a formal theory that is different from the symbolic-algorithmic theory of today's computer science textbooks. We propose a general strategy for developing such a theory, and within that general view, a specific approach that we call "fluent computing". In contrast to Turing, who modeled computing processes from a top-down perspective as symbolic reasoning, we adopt the scientific paradigm of physics and model physical computing systems bottom-up by formalizing what can ultimately be measured in any physical substrate. This leads to an understanding of computing as the structuring of processes, while classical models of computing systems describe the processing of structures.

@thezerobit for fielded devices I gotta agree. However, for algorithmic explanation of an approach to computing that might *actually* be well suited to memristors (and other proposed noisy hardware technologies) there's a reasonable amount of activity in the Vector Symbolic Architecture / Hyperdimensional Computing space:

duckduckgo.com/?t=h_&q=memrist

duckduckgo.com/?t=h_&q=memrist

duckduckgo.commemristor "vector symbolic architecture" at DuckDuckGoDuckDuckGo. Privacy, Simplified.

"#HyperdimensionalComputing: Each piece of information is represented as a single entity, a hyperdimensional vector. An advantage of HC is transparency: The algebra clearly tells you why the system chose the answer it did. The same is not true for traditional neural networks. HC is well suited for low-power hardware. Despite such advantages, HC is still in its infancy but there’s real potential here."

A New Approach to Computation Reimagines Artificial Intelligence
quantamagazine.org/a-new-appro

Quanta MagazineA New Approach to Computation Reimagines Artificial Intelligence | Quanta MagazineBy imbuing enormous vectors with semantic meaning, we can get machines to reason more abstractly — and efficiently — than before.
Replied in thread

@DrYohanJohn
@cogsci

Boosting for my followers - see original post in the thread above.

The BBS target article on the Language of Thought Hypothesis states:

"We outline six core properties of
LoTs: (i) discrete constituents; (ii) role-filler independence; (iii) predicate-argument
structure; (iv) logical operators; (v) inferential promiscuity; and (vi) abstract content."

It's interesting to think about the extent to which those properties are directly enabled by neural representations using Vector Symbolic Architectures / Hyperdimensional Computing. It hadn't occurred to me to draw a line between LoT and VSA/HDC.