

In [1]:

    
%matplotlib inline
from abm import analysis, nxpops, io



In [2]:

    
%pylab inline
pylab.rcParams['figure.figsize'] = (10, 6)









    



Populating the interactive namespace from numpy and matplotlib



In [3]:

    
cfg = io.ConfigReader('../setup.json').get_config()
smpop = nxpops.SoftmaxNxEnvironment(**cfg)
df = analysis.get_env_likelihood_samples(smpop, n_tasks=4800, sample_each=1200)

Learning Plots

Observe that the likelihood of the shortest path increases!



In [4]:

    
segments = analysis.plot_learning_df(df, key='li', no_mismatch=False, full_mismatch=False, some_mismatch=True)

Learnt vs Shortest Path

And observe that the most-probable learned path is now a smaller factor longer than the true shortest



In [11]:

    
analysis.plot_learning_df(df, key='learnt_over_best',
                          no_mismatch=False, full_mismatch=False, some_mismatch=True)



In [ ]: