Parameter analysis

January 2021

Cross validation by top sorting: Classification accuracy and zero entry count

heatmap-top-cv


Cross validation by bottom sorting: Classification accuracy and zero entry count

heatmap-bottom-cv


Cross validation: Difference between top and bottom

heatmap-difference-cv

50 random chiefs for selection (average over 20 trials)

random1

Same data, each trial presented individually

heatmap-random-cv

Activity data randomly permuted (average over 20 trials)

random4

Distribution of parameter values in circuit

densityplot


Top 50 chief analysis by two parameters

densityplot-2x-top50

Classification accuracy databases used to make images on this page

  • classification_accuracies.pkl: Pandas dataframe of classification accuracies of top parameter values (cv, cv_error, test, test_error)
  • zeros.pkl: Pandas dataframe of number of zero entries in top feature vectors (out of a total of 450056 entries)
  • lowclassification_accuracies.pkl: Pandas dataframe of classification accuracies of bottom parameter values (cv, cv_error, test, test_error)
  • lowzeros.pkl: Pandas dataframe of number of zero entries in bottom feature vectors (out of a total of 450056 entries)
  • randomclassification_accuracies.pkl: Pandas dataframe of classification accuracies of random parameter values (cv, cv_error, test, test_error)
  • randomzeros.pkl: Pandas dataframe of number of zero entries in random feature vectors (out of a total of 450056 entries)
Main dataframe and Python code