alan-turing-institute/tapas

Issues

• Code example for privacy attacks with completely unknown generator

  #127 opened a year ago by tiadams
  1

• Improve the reporting function (especially when disaggregating by target_id), making the labels etc neater.

  #115 opened a year ago by fhoussiau
  0

• Confidence intervals in evaluation.

  #129 opened a year ago by fhoussiau
  0

• Remove targets from dataset

  #122 opened a year ago by fhoussiau
  0

• QOL improvements in generation

  #124 opened a year ago by fhoussiau
  0

• Allow creation of dataset schema from dataframe/csv

  #105 opened 2 years ago by fhoussiau
  0

• Get errors in quickstart

  #123 opened a year ago by f-hafner
  2

• generate_training_samples always produces all samples

  #126 opened a year ago by fhoussiau
  0

• Rename toolbox, add link to paper, add code for paper

  #108 opened a year ago by fhoussiau
  0

• Literature review issue (keep open)

  #75 opened 2 years ago by fhoussiau
  0

• Add a "canary in the coalmine" example

  #116 opened 2 years ago by fhoussiau
  0

• Add examples *from the paper* in an `examples` folder for easy explanation of how the code works.

  #70 opened 2 years ago by fhoussiau
  2

• Enable threat models that use only datasets as inputs.

  #113 opened 2 years ago by fhoussiau
  0

• Support for no-box attacks.

  #112 opened 2 years ago by fhoussiau
  0

• Implement ImpreciseBox attacks (generators)

  #111 opened 2 years ago by fhoussiau
  0

• Error package TAPAS

  #110 opened 2 years ago by FerdoosHN
  0

• Enable the user to specify whether to re-randomise the sensitive attribute in AIAs.

  #106 opened 2 years ago by fhoussiau
  1

• ROC curve reporting: disaggregate by variable.

  #104 opened 2 years ago by fhoussiau
  0

• Implement Zanella et al. (https://arxiv.org/pdf/2206.05199.pdf) for effective epsilon estimation.

  #103 opened 2 years ago by fhoussiau
  0

• Formatting of reports (nice plots)

  #102 opened 2 years ago by fhoussiau
  0

• Handling seeds in generators

  #101 opened 2 years ago by fhoussiau
  0

• Fix tests

  #99 opened 2 years ago by fhoussiau
  0

• Add `.scores` to `MIAttackSummary`.

  #89 opened 2 years ago by fhoussiau
  1

• Add reporting for AIAs, especially in the more general case (>2 classes).

  #91 opened 2 years ago by fhoussiau
  1

• Save/Load threat models

  #87 opened 2 years ago by fhoussiau
  0

• Add MultiplyTargetedMIA, a "meta" MIA that allows to instantiate multiple TargetedMIA.

  #88 opened 2 years ago by fhoussiau
  0

• Add simple attacks

  #86 opened 2 years ago by fhoussiau
  0

• Bug in sample

  #90 opened 2 years ago by fhoussiau
  0

• Simplify `Report`

  #77 opened 2 years ago by fhoussiau
  0

• Investigate whether it makes sense to have `TargetedMIA.test` output a `MIAttackSummary` directly.

  #74 opened 2 years ago by fhoussiau
  0

• Modify the `AuxiliaryDataKnowledge` to split the auxiliary dataset, rather than having to do it manually

  #84 opened 2 years ago by fhoussiau
  0

• Complete the implementation of the Groundhog attack

  #73 opened 2 years ago by fhoussiau
  0

• Add `tqdm` support

  #71 opened 2 years ago by fhoussiau
  0

• Extend Generator class to allow passing config

  #67 opened 2 years ago by callummole
  0

• Add example of using reprosyn from the generator class

  #63 opened 2 years ago by callummole
  0

• Change `write_to_string()` and `_parse_cv()` to allow for headered data

  #83 opened 2 years ago by callummole
  1

• Remove this line?

  #82 opened 2 years ago by andeElliott
  1

• Json structure

  #80 opened 2 years ago by andeElliott
  0

• Speed up

  #79 opened 2 years ago by andeElliott
  0

• Read this paper (Bayesian estimation of DP)

  #78 opened 2 years ago by fhoussiau
  0

• Add config support for exe (within reprosyn)

  #50 opened 2 years ago by JamesJordon
  0

• Add Attribute-inference attacks as a threat-model, and in the groundhog attack.

  #55 opened 2 years ago by fhoussiau
  0

• Add contributors file

  #66 opened 2 years ago by andeElliott
  0

• Issue with prive command line tool

  #60 opened 2 years ago by andeElliott
  2

• Prive command line tool - default config file

  #65 opened 2 years ago by andeElliott
  0

• Implement effective $\varepsilon$ as a metric.

  #58 opened 2 years ago by fhoussiau
  0

• Add worst-case attack model for MIA/AIA (knowing everything except the target bit).

  #57 opened 2 years ago by fhoussiau
  0

• Add dryrun to generator class

  #56 opened 2 years ago by callummole
  0

• Enable running of prive and rsyn from command-line

  #53 opened 2 years ago by triangle-man
  1

• implement ClosestDistanceAttack

  #51 opened 2 years ago by fhoussiau
  0