Quick and Dirty Intro to Neurosymbolic AI

The Neuro Symbolic Channel provides the tutorials, courses, and research results on one of the most exciting areas in artificial intelligence and machine learning. With content originally from the AI course taught at Arizona State University, this channel brings you the latest at the intersection of symbolic methods (e.g., logic programming) and deep learning. Learn about the latest algorithms, Python packages, and progress toward larger goals such as artificial general intelligence (AGI).

[@HardwarizeVault]

We need more videos like this one. Neurosymbolic AI is still a not so known field and is difficult for begineers to starts without this kind of introductory materials, tutorials...

[@IdPreferNot1]

Just discovered your channel.. Look forward to seeing more as codifying reasoning seems like the area LLMs need the most help with to improve our agentic systems.

[@MrAndreaCaso]

I've been waiting for something like this, thank you so much for posting :-) !

[@existenceisillusion6528]

On a side note ... Old Ironsides
Also, looking forward to learning more about NSAI

[@pabloe1802]

any book, course to recommend?

[@Isaacmellojr]

Does where come from the assumption that humans does not need "extensive datasets" to do some razoning task? How much neurobytes do you need to recognize an object a face, to say "dada".. and to start to walk. To move your arms. You need years of multysensory input . This is a hudge dataset. Isn't?

[@Isaacmellojr]

"Simbolic AI does not need trainig data" this sentence does make any sense. Simbolic AI already deal with predefined "ALREADY KNOW" patterns.

[@MuantanamoMobile]

*reading and listening comprehension

[@Isaacmellojr]

@@MuantanamoMobile after check transciption. He truly say that SIMBOLIC AI does not need training data. It make sense. But simbolic AI do not need training data because it deal with pre define concept. While deep learning deals with pattern recognition.
It's possible I make some mistake, but u show no evidence of that.

[@Isaacmellojr]

@@MuantanamoMobile 14:12

[@MuantanamoMobile]

@@Isaacmellojr "Does not need training data SPECIFIC for EACH combinatorial outcome" - Do you know what this means?
It doesn't mean NEURO - SYMBOLIC AI doesn't need training data, rather it doesn't need to have data on every *single possible example* in order to give accurate results, unlike machine learning like DL.

[@MuantanamoMobile]

@@Isaacmellojr This means: Neuro-symbolic AI doesn't need to see a picture of bird or dog, in every possible situation, or pose in order to accurately infer that it is bird dog.
While ML does need more and more data (of different pictures of birds and dogs) up to a point, however it later hits a bottleneck - and more data can then lead to paradoxical problems like overfitting and underfitting etc.