I'm looking for a neural network model with specific characteristics. This model may not exist...

I need a network which doesn't use "layers" as traditional artificial neural networks do. Instead, I want [what I believe to be] a more biological model.

This model will house a large cluster of interconnected neurons, like the image below. A few neurons (at bottom of diagram) will receive input signals, and a cascade effect will cause successive, connected neurons to possibly fire depending on signal strength and connection weight. This is nothing new, but, there are no explicit layers...just more and more distant, indirect connections.

As you can see, I also have the network divided into sections (circles). Each circle represents a semantic domain (a linguistics concept) which is the core information surrounding a concept; essentially a semantic domain is a concept.

Connections between nodes within a section have higher weights than connections between nodes of different sections. So the nodes for "car" are more connected to one another than nodes connecting "English" to "car". Thus, when a neuron in a single section fires (is activated), it is likely that the entire (or most of) the section will also be activated.

All in all, I need output patterns to be used as input for further output, and so on. A cascade effect is what I am after.

I hope this makes sense. Please ask for clarification where needed.

Are there any suitable models in existence that model what I've described, already?

enter image description here

  • $\begingroup$ it's not really clear to me what the goal of this neural network is, what the distinction between "layer" and the circles in your diagram, or why this is "more biological". If you are interested in neural networks of semantic cognition, there are several. Perhaps you could start here: psych.stanford.edu/~jlm/papers/McCRogers03.pdf $\endgroup$ – Jeff Nov 14 '12 at 23:53
  • $\begingroup$ Awesome, I'll have a look. The goal of this network is to translate a word from English to Spanish. It does so by separating storage of words from concepts. The concept of a car is used by both English and Spanish. Given that we start in English and want to end with Spanish, the network will find the Spanish word corresponding to the car concept. $\endgroup$ – Chad Johnson Nov 15 '12 at 0:45
  • 1
    $\begingroup$ ah interesting! keep in mind there is a lot of literature on bilingualism and translation out there; there is at least some evidence to suggest translating does not need to activate semantic features (perhaps two lexical nodes are joined at the lemma level). $\endgroup$ – Jeff Nov 15 '12 at 1:08
  • 1
    $\begingroup$ yes, basically. most models of spoken word production posit at least 3 levels: semantic features, lemma, and phonemes. if words of different language share the same lemma (this is debated), then translating a word may involve going up only one level, from phonemes to lemma, and then back down to the phonemes of another language, without accessing the semantic features of a word $\endgroup$ – Jeff Nov 15 '12 at 19:01
  • 1
    $\begingroup$ This is a naive view of translation. Translation is much, much more complicated than a 1 to 1 mapping between words... even though the economic stakes are huge, we are still very, very far from machine translation. (Douglas Hofstadter has written some interesting books on the difficulty of translation, if you are interested). $\endgroup$ – Timothée Behra Sep 9 '13 at 14:17