Quantitative Evaluation of the Expected Antagonism of Explainability and Privacy: Unterschied zwischen den Versionen

Aus SDQ-Institutsseminar
Keine Bearbeitungszusammenfassung
Keine Bearbeitungszusammenfassung
 
(2 dazwischenliegende Versionen von 2 Benutzern werden nicht angezeigt)
Zeile 2: Zeile 2:
|vortragender=Martin Lange
|vortragender=Martin Lange
|email=martin.lange@student.kit.edu
|email=martin.lange@student.kit.edu
|vortragstyp=Proposal
|vortragstyp=Bachelorarbeit
|betreuer=Clemens Müssener
|betreuer=Clemens Müssener
|termin=Institutsseminar/2021-06-11
|termin=Institutsseminar/2021-08-20
|kurzfassung=Explainers for machine learning models help humans and models work together. They build trust in a model's decision by giving further insight into the decision making process. However, it is unclear whether this insight can also expose private information. The question of our thesis is whether there exists a conflict of objectives between explainability and privacy and how we measure the effects of this conflict. Specifically we are looking at local feature importance explainers.
|kurzfassung=Explainable artificial intelligence (XAI) offers a reasoning behind a model's behavior.
 
For many explainers this proposed reasoning gives us more information about
We propose a use case where the prediction of a model for a person is considered their private data. An attacker might be able to gain insight into the predictions for other people by abusing their own explanation to imitate the model's behavior. We will test this use case experimentally to determine whether such an attack is possible.
the inner workings of the model or even about the training data. Since data privacy is  
becoming an important issue the question arises whether explainers can leak private data.
It is unclear what private data can be obtained from different kinds of explanation.
In this thesis I adapt three privacy attacks in machine learning to the field of XAI:
model extraction, membership inference and training data extraction.  
The different kinds of explainers are sorted into these categories argumentatively and I present specific use cases how an attacker can obtain private data from an
explanation. I demonstrate membership inference and training data extraction for two specific explainers in experiments. Thus, privacy can be breached with the help of explainers.
}}
}}

Aktuelle Version vom 10. August 2021, 12:16 Uhr

Vortragende(r) Martin Lange
Vortragstyp Bachelorarbeit
Betreuer(in) Clemens Müssener
Termin Fr 20. August 2021
Vortragsmodus
Kurzfassung Explainable artificial intelligence (XAI) offers a reasoning behind a model's behavior.

For many explainers this proposed reasoning gives us more information about the inner workings of the model or even about the training data. Since data privacy is becoming an important issue the question arises whether explainers can leak private data. It is unclear what private data can be obtained from different kinds of explanation. In this thesis I adapt three privacy attacks in machine learning to the field of XAI: model extraction, membership inference and training data extraction. The different kinds of explainers are sorted into these categories argumentatively and I present specific use cases how an attacker can obtain private data from an explanation. I demonstrate membership inference and training data extraction for two specific explainers in experiments. Thus, privacy can be breached with the help of explainers.