Trustworthy Machine Learning

Trustworthy Machine Learning

Machine learning has made great advances over the past year and many techniques have found their ways into applications. This leads to an increasing demand of techniques that not only perform well - but are also "trustworthy".

Trustworthiness includes:

• Interpretability of the prediction
• Robustness against changes to the input, which occur naturally or with malicious intend
• Privacy preserving machine learning (e.g. when dealing with sensitive data such as in health applications)
• Fairness
• Model Stealing
• Interprability
• Authenticity 

 

Schedule

• 2021 11 05: Kickoff Meeting (slides)
• 2021 11 12: How to (slides)
• 2021 12 03 Round 1 Group 1:
  • Privacy, Model Stealing,
• 2021 12 10 Round 1 Group 2:
  • Uncertainty, Interpretability, Fairness, Authenticity
• 2022 01 28 Round 2 Group 1:
  • Privacy, Model Stealing, Authenticity
• 2022 02 04 Round 2 Group 2:
  • Uncertainty, Interpretability, Fairness

 

Paper list and topic assignments:

 

Uncertainty (David)
Apratim Bhattacharyya; Mario Fritz; Bernt Schiele
Long-Term On-Board Prediction of People in Traffic Scenes under Uncertainty Inproceedings
In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018

Model Stealing (Niklas)
Tribhuvanesh Orekondy; Bernt Schiele; Mario Fritz
Knockoff Nets: Stealing Functionality of Black-Box Models 
In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019

Interpretability (Robert)
Moritz Bohle, Mario Fritz, Bernt Schiele
Convolutional Dynamic Alignment Networks for Interpretable Classifications 
In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2021

Robustness to Adversarial Examples
Seong Joon Oh; Mario Fritz; Bernt Schiele
Adversarial Image Perturbation for Privacy Protection -- A Game Theory Perspective 
In: IEEE International Conference on Computer Vision (ICCV), 2017

Authenticity / Deep Fakes (Krisztian)
Ning Yu; Larry Davis; Mario Fritz
Attributing Fake Images to GANs: Learning and Analyzing GAN Fingerprints 
In: International Conference on Computer Vision (ICCV), 2019

Privacy (Ady)
Tribhuvanesh Orekondy; Bernt Schiele; Mario Fritz
Towards a Visual Privacy Advisor: Understanding and Predicting Privacy Risks in Images 
In: IEEE International Conference on Computer Vision (ICCV), 2017

Fairness (Anton)
Ning Yu; Ke Li; Peng Zhou; Jitendra Malik; Larry S. Davis; Mario Fritz
Inclusive GAN: Improving Data and Minority Coverage in Generative Models 
In: European Conference on Computer Vision (ECCV), 2020

Causality
Vedika Agarwal; Rakshith Shetty; Mario Fritz
Towards Causal VQA: Revealing and Reducing Spurious Correlations by Invariant and Covariant Semantic Editing
In: IEEE Conference on Computer Vision and Pattern Recognition, 2020.
 

 

Description

As a proseminar’s primary purpose is to learn presentation skills, the seminar will feature two presentations from each student. 

In the first half of the semester, we will have presentations of two topics each week. After each presentation, fellow students and lecturers will provide feedback on how to improve the presentation. This general feedback must then be taken into account for the second half of the semester, where again each student will present.

 

Grading

The first presentations  will count towards 30% of the overall grade, the second presentation will count towards 70% of the overall grade. Attendance in the proseminar meetings is mandatory. At most one session can be skipped, after that you need to bring a doctor’s note to excuse your absence.

 

 

Zoom Link (trying to offer hybrid setting for presentation - update to follow)

Proseminar: Trustworthy ML
Location: https://cispa-de.zoom.us/j/96664715917?pwd=YjJwRWJEamZrZ3RtWmRtZG1NaHNvZz09
Mario Fritz is inviting you to a scheduled Zoom meeting.

Topic: Proseminar: Trustworthy ML
Time: This is a recurring meeting Meet anytime

Join Zoom Meeting
https://cispa-de.zoom.us/j/96664715917?pwd=YjJwRWJEamZrZ3RtWmRtZG1NaHNvZz09

Meeting ID: 966 6471 5917
Passcode: ^CL3?x
One tap mobile
+496971049922,,96664715917# Germany
+496950502596,,96664715917# Germany

Dial by your location
        +49 69 7104 9922 Germany
        +49 695 050 2596 Germany
Meeting ID: 966 6471 5917
Find your local number: https://cispa-de.zoom.us/u/acDUElCccb

Privacy Notice:
Please refer to https://cispa.de/en/data-privacy-policy-zoom for our privacy notice regarding the use of Zoom at CISPA