an eelnet made by man for the eel fighting: An introduction to network theory and practice
The title of this entry contains the penultimate line in a poem by Robert Lowell entitled Dolphin. The net that binds the confessional poet has many strands: the iron-clad spoils of his pitiless memory, the effervescence of his creative talent, the injury he has done to others and to himself, and guilt, terrible guilt in the unforgiving final line: my eyes have seen what my hand did.
In Hindu and Buddhist teachings, sanskaras (variously translated as activities, conditions, discriminations, fabrications, formations or impressions, but also fibers or threads), have served for millennia in descriptions of human sorrow and prescriptions for its abatement. Human beings, their minds conditioned by ignorance of the truth, spin nets or webs of suffering around themselves by their own thoughts and actions.
The modern Indian spiritual teacher, Meher Baba, writes extensively about the form and function of sanskaric threads. In his DISCOURSES he notes, The mental processes are partly dependent upon the immediately given objective situation, and partly dependent upon the functioning of accumulated sanskaras or impressions of previous experience….
So it is that the vivid trope of nets, now networks, has been an image in the collective human mind for a very long time, has been re-tooled for human use throughout history and beyond history. And of course, the internet, the materialization of Teilhard de Chardin’s noosphere.
Perhaps because of their archetypal status, the mathematics of networks seems to many people more familiar, and easier to grasp.
A lot of good neuroscience comes out of British Columbia, as well as excellent pedagogy. Today’s site – AI space (aispace.org) – is an interactive tutorial on neural networks, built by the good folks at the Laboratory for Computational Intelligence at the University of British Columbia.
It’s true there are lots of big-ticket programs that you want to use for calculating your neural network data, but these have steep learning curves and are not set up to teach you the basics. The designers at AI space provide Java applets for a wide range of topics. Within the applet, you can design, analyze and solve a particular neural net architecture. Topics include: graph searching, constraint satisfaction problems, deduction, belief networks, and neural nets. Instructions include written and video, step-by-step recipes.