fibber.benchmark.benchmark_utils module¶
-
fibber.benchmark.benchmark_utils.load_detailed_result()[source]¶ Read detailed results from file.
- Returns
the detailed result table. Returns an empty DataFrame if file does not exist.
- Return type
(pandas.DataFrame)
-
fibber.benchmark.benchmark_utils.reorder_columns(results)[source]¶ Reorder columns in the result table.
- Parameters
results (DataFrame) – a result table
- Returns
DataFrame
-
fibber.benchmark.benchmark_utils.update_attack_robust_result(aggregated_result, robust_tuned_clf_desc, robust_tuning_steps, result_dir=None)[source]¶ Read results of attacking robust classifiers, and add a row to the file. Create a new file if the table does not exist.
- Parameters
aggregated_result (dict) – the aggregated result as a dict.
robust_tuned_clf_desc (str) – the robust tuning description.
robust_tuning_steps (int) – the number of robust tuning steps.
result_dir (str or None) – the directory to save results. If None, use
<fibber_root_dir>/results/.
-
fibber.benchmark.benchmark_utils.update_detailed_result(aggregated_result, result_dir=None)[source]¶ Read dataset detailed results and add a row to the file. Create a new file if the table does not exist.
- Parameters
aggregated_result (dict) – the aggregated result as a dict.
result_dir (str or None) – the directory to save results. If None, use
<fibber_root_dir>/results/.