aimclub / GOLEM

Graph Optimiser for Learning and Evolution of Models

Home Page:https://thegolem.readthedocs.io

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Two stages of graph evaluating

kasyanovse opened this issue · comments

commented

Graph evaluating consumes a lot of time because it does on all data. If it is used for graph verification it is no need for running on full dataset, it may be done on tiny data part. Therefore there may be two stage of evaluating:

  1. Evaluating as verification on tiny dataset
  2. Evaluating for fitness calculation on full dataset

There are two way to do it:

  1. Get two different evaluators for graph composing
  2. Add interface for control size of dataset is used for evaluation