Consider removing low probability attribute values from nodes to speed up calculations (less attribute values to compute over).

cmaclell / concept_formation

Python implementations of TRESTLE, COBWEB/3, and COBWEB

MIT License

61 stars 18 forks source link

Consider removing low probability attribute values from nodes to speed up calculations (less attribute values to compute over). #39

Open cmaclell opened 7 years ago

cmaclell commented 7 years ago

The idea would be that when attribute values dip below a certain probability then they get removed. Not sure on any more specifics, but wanted to put the idea on the list.