Dataflow matrix machines as a bridge between programs and neural nets

[anhinga_anhinga]

Continuing the last couple of posts

https://anhinga-anhinga.livejournal.com/82953.html

and

Comments

[cema]

I think Lisp had associative lists, as an approximation of dictionaries. (Linear search time though.)

[anhinga_anhinga]

It was also an afterthought, not something the language designers were proud of... Not something one was encouraged to use on a daily basis...

***

But I think the reason we are using lists in Lisps is that Lisp is 60 years old. Whereas, if it were 50 years old, it's quite likely we would be using dictionaries.

See e.g. something like Vega or Vega-Lite, which just use JSON to declaratively define interactive programs (in a limited domain):

https://vega.github.io/vega-lite/

[anhinga_drafts]

We finally implemented a bit of "editing the network on the fly" schema last week:

https://github.com/jsa-aerial/DMM/tree/master/examples/dmm/quil-controlled/interactive