3 ms·
There is a dozen of papers in this methodology (e.g. end of https://community.wolfram.com/groups/-/m/t/3017754 https://community.wolfram.com/groups/-/m/t/301775
by jarekd 2y ago
There is a dozen of papers in this methodology (e.g. end of https://community.wolfram.com/groups/-/m/t/3017754 https://community.wolfram.com/groups/-/m/t/3017754 ), but not as ANN.
However, it degenerates to ~KAN if restring to pairwise dependencies (can consciously add triplewise and higher), and gives many new possibilities, like multidirectional propagation, of values or probability distributions, with novel additional training approaches like through tensor decomposition.
- cs702 2y agoWhen I ask if this has been tested, I meant as an ANN on conventional benchmarks. Sorry if that wasn't clear. There are a lot of ideas that are clever and seem promising... but fail to perform well on such benchmarks. Is there a github repo with code available?
- jarekd 2y agoWhich benchmarks for multidirectional neurons? To compare with which approaches? Multidirectional are biological neurons, but I don't know how to compare with them?
- cs702 2y agoCan you show the world this can be made to work for, say, a toy benchmark like MNIST classification? --- To be 100% clear: My question about practical application today is orthogonal to the question about whether this research is worth pursuing!
- jarekd 2y ago(Multidirectional) biological neural networks are no longer superior in MNIST benchmark ... but e.g. consciousness, or being able to learn from single examples. And no, recreating it is not a task a single person can complete.
- cs702 2y agoAlright. I've added you preprint to my reading list, so I can take a closer look at this.
- jarekd 2y agoJust represent joint density for each neuron as a linear combination - then you can inexpensively propagate in both directions e.g. as E[X|Y,Z] or E[Y,Z|X] by substituting and normalizing ... the formulas turn out quite simple - could be hidden in dynamics of (bidirectional) biological NN ... And for pairwise distribution becomes ~KAN, which turned out quit successful ... so we are talking about its extension: adding more possibilities, like triplewise dependencies and multidirectional propagation.