Hey, I’ve realized the best category for me is Neuro-symbolic AI.
I thought to say hello and welcome messaging on the topic.
-Ernst03
Welcome! Neuro-symbolic AI is a good area to explore because it combines two strengths: neural models are strong at learning patterns from data, while symbolic methods are strong at explicit rules, logic, and structured knowledge. That combination can be especially useful when a result needs to be checked, explained, or kept within clear constraints. It would be interesting to hear which side interests you more, such as reasoning, knowledge graphs, or building a small project that uses both approaches.
Yes, I am learning that Neuro-Symbolic AI is an area where my experience seems particularly well suited.
I am retired now, and over the years, long before modern AI, I was exploring binary transformations and mathematical structures that I now recognize as finite dynamical systems.
I want to learn how to engineer AI systems. I have a ways to go, but I also bring more than 35 years of experience working with binary constructs to the table.
I believe what I’ve learned may have a place in AI, particularly in how information is represented and processed.
There is much I don’t yet know about signals, systems, and neural networks, but we all start somewhere. I figured finding friends who share these interests would be a good step.
One area that particularly interests me is finite binary dynamical objects, including periodic orbits under defined transformations. I’ve been exploring these structures for years, and I’m interested in how they might be used in dynamical computational systems.
I’m particularly interested in the intersection of symbolic representation and dynamical processing, and I hope to learn how these ideas might fit into Neuro-Symbolic AI.
I am reading that researchers are very interested in understanding what happens inside AI systems between the prompt and the output. We know the mathematical operations, but understanding how the internal representations produce particular results remains a challenge.
I’ve been wondering whether a common binary encoding system, used from input through processing to output, might provide a foundation for a different kind of AI architecture. Perhaps one common mathematical language for all its functions.
Does the challenge of understanding AI’s internal processing interest you, @fabbankbalnancecheck?
-Ernst03
Hey, I am now on Discord.
I think that is what works best.
I have been hanging around these main forums in Research mostly but I am serious about my projects and Neuro-Symbolic “signals and systems” does seem to be the right direction going forward.
So i assume the correct usage of these forums is more question and answer over a continuous chat form.
So I am Ernst03 or ernst_berg on Discord.
I’ll move my “chats” there so I show respect for Hugging Face forums.
Wow! It takes a bit of work to get everything set up, but I’ve created a Discord server!
Experimental Artificial Intelligence (EAI)
The idea is to provide a home for people like myself who want to explore experimental AI, develop projects, and share their work with a community of like-minded people.
Members are welcome to have their own project channels, exchange ideas, collaborate, or simply enjoy the conversation. All experience levels are welcome.
Discord has its own rules about sharing invitations, so for now I’m distributing the link manually. Feel free to pass it along to anyone who might be interested!
If you just want to chat, that’s perfectly fine too. We have a #general channel for that.
I’m retired, and I expect to spend the rest of my life exploring the things that interest me. I’m quite happy spending my days working with computers, mathematics, and AI. But having a place to share that experience, make friends, and participate in a community is important too.
That’s really the inspiration behind EAI. I’d like a place where we can enjoy ongoing conversations without cluttering up the Hugging Face Research or Beginners forums, which serve somewhat different purposes.
As I’ve mentioned, I’m a senior, and I’ll do my best to be a fair and welcoming moderator. Someone has to own the server, and I understand that comes with some responsibility!
I’ve started a Dynamic Unary channel to share my own research. Dynamic Unary involves what I think of as mathematical objects of the dynamical kind — finite periodic orbits that may offer interesting ways to represent and process information. One of my ambitions is to explore whether these constructs might be useful in designing alternative neural networks or other computational architectures.
But EAI isn’t intended to be just about my work. I hope others will bring their own ideas, experiments, and projects.
There is so much involved in artificial intelligence today that I sometimes wonder whether anyone can really be a polymath in the field anymore!
Perhaps that’s another reason to build a community. None of us needs to know everything if we’re willing to learn from one another.
Bring an idea, share an experiment, or just come make a friend.
— Ernst