hopefully, I'll post some articles on some networks which are
random,
dynamic,
plasitic,
stochastic.
For randomness of network, I have such an idea like this.
Here are 2 agents 'A' and 'B', and 1 key-value database.
'A' puts some data to the database.
'B' gets the data from the database.
The relationship of 'put'-'get' for 'A' and 'B' can be a 'network'.
When so many agents are out there, 'put'-'get' relation (network linkage) would not be stable. Sometime 'A' puts, and 'C' gets. or 'B' gets again.
If 'A'-'B' linkage were stable, there would be some reason.
1) 'A' and 'B' share almost same timing of 'get', 'put' operation.
2) Only 'A' and 'B' are connected the same database.
Then, I'd like to define 'distance' or 'closeness' on the "network" of 'A' and 'B' nodes.
For 1) case, time dimension may be used to measure the 'distance'.
For 2) case, connection (physical distance) may be used to measure the 'distance'.
So, I modeled some "random" network and distance (inverse of linkage strong).
Then I'd like to expand them to the limit.
see http://www.cis.twcu.ac.jp/~asakawa/waseda2002/elman.pdf
ReplyDeleteand see more http://d.hatena.ne.jp/m-a-o/20130917#p2
ReplyDeleterecurrent neural network
ReplyDeleteHopfield network
Boltzmann machine
coherent state物語
http://d.hatena.ne.jp/m-a-o/20130303#p2
http://en.wikipedia.org/wiki/Earth_Mover%27s_Distance
ReplyDeleteabove link came from http://aidiary.hatenablog.com/entry/20120804/1344058475
Delete