Exploring Multidimensional Neurons: A Novice's Perspective

Hello OpenAI Community.

I’m an enthusiast fascinated by the potential of artificial intelligence, and I’ve been pondering a concept related to neural networks. I’d like to share my thoughts and get feedback from this knowledgeable community, even though I have no formal expertise in the field. Please bear with me as I try to articulate my idea.

I’ve been thinking about the traditional neural network model and wondered about the possibilities of using multidimensional neurons instead of conventional weights. My understanding is based on casual exploration and discussions with chatGPT.

Here are some Key points:

  1. Spatial Mapping: Considering the spatial arrangement of neurons.
  2. Transformer-Based Training: Exploring the role of a Transformer-based Trainer in guiding network creation.
  3. Adaptability: How can multidimensional neurons adapt to different problem domains?

I’d love to hear your thoughts, insights, and critiques regarding this concept.

1 Like

Maybe look into neuromorphic computing.

One place to start is the nanowire networks:


Or also the new DeepSouth computer capable of 228 trillion synaptic ops per second:


Thank you for sharing this information. It’s fascinating to see the progress in neuromorphic computing. The parallels between the principles described in the article, such as parallel processing and energy efficiency inspired by the human brain, and the concepts I’ve been exploring in my abstract are striking. It reinforces the idea that the spatial mapping of neurons and the emulation of biological processes hold significant promise for advancing both computing capabilities and our understanding of neurological systems. The convergence of these efforts, exemplified by projects like DeepSouth, indeed seems to be a compelling validation of the abstract’s premise.

Hello. You might also find interesting:

  1. Biological neuron model
  2. Single cortical neurons as deep artificial neural networks
  3. Cellular model
  4. Differentiable self-organizing systems