tanjascats/fingerprinting-toolbox

Fingerprinting toolbox that allows embedding and extracting fingerprints into the relational data

fingerprinting-toolbox's Issues

float attributes raise a TypeError

When trying the universal scheme insertion (fingerprinted_data_weighted = scheme.insertion(main_X, secret_key=secret_key, recipient_id=recipient_id, attributes_weights=rf_model.feature_importances_)) I get the following error:

Traceback (most recent call last):
  File "run.py", line 240, in <module>
    main()
  File "run.py", line 181, in main
    fingerprinted_data_weighted = scheme.insertion(main_X, secret_key=secret_key, recipient_id=recipient_id, attributes_weights=rf_model.feature_importances_)
  File "~/fingerprinting-toolbox/scheme/_universal.py", line 158, in insertion
    marked_attribute = set_bit(attribute_val, bit_idx, mark_bit)
  File "~/fingerprinting-toolbox/utils.py", line 87, in set_bit
    val = val & mask
TypeError: unsupported operand type(s) for &: 'numpy.float64' and 'int'

To reproduce the error use the Bank_Personal_Loan_Modelling.csv dataset (https://www.kaggle.com/krantiswalke/bank-personal-loan-modelling-supervised-learning). The problem happens exactly with the 'CCAvg' attribute (type float) and is resolved if the latter is excluded.