Workshops

Track I: Interpretability & Explainable AI (Zeynep Akata, Eric Raidl, Lenon Minorics)

Venue: MPI-IS Lecture Hall

Chair: Valkyrie Felso

The workshop will be organized as a series of talks by Zeynep Akata, Eric Raidl and Lenon Minorics with time for discussions after each talk.

• 1.00 - 1:55 Eric Raidl: Explanation, Explainability and Conditionals
• 1:55 - 2:50 Zeynep Akata: Explainable Machine Learning with Natural Language
• 2.50 - 3.20 Coffee Break (MPI-IS Lobby)
• 3.15 - 4.15 Lenon Minorics: A Game Theoretical Approach to Explainable AI

Track II: Robust Action Learning (Georg Martius & Sebastian Blaes)

Venue: Max Planck House, Lecture Hall

Chair: Indu Ilanchezian

• 1.00 - 1.15 Opening — Georg Martius (AL Group)
• 1.15 - 1.40 Control what you can — Sebastian Blaes (AL Group)
• 1.40 - 2.05 Unleashing RL onto Real Robots through Robustness — Dieter Büchler (EI Department)
• 2.05 - 2.30 Robustness and conservativeness in decision making – an optimization perspective — Jia-Jie Zhu (EI Department)
• 2.30 - 2.45 Discussion
• 2.45 - 3.15 Coffee Break in the MPI-IS Lobby
• 3.15 - 4.30 Hands-on: Robust Policy Distillation with DAgger — Georg Martius and Sebastian Blaes (AL Group)

Important Note for Participants of Track II: Please bring your laptops for this workshop.

Link to Course Material: https://owncloud.tuebingen.mpg.de/index.php/s/r3syDbH76WJ3QFn

• Note: You need a MuJoCo license which you can obtain here: http://www.mujoco.org/index.html

Track III: Neuroscience & ML (Dylan Paiton, Santiago Cadena)

Venue: MPI for Biological Cybernetics, Seminar Room

Chair: Mohammad Bashiri

Dylan Paiton and Santiago Cadena (Bethge Lab, Uni Tübingen) host this session of talks about topics in neuroscience and ML.

• 1.00 - 1.05 Welcome & Intro (Mohammad Bashiri)
• 1.05 - 1.40 Santiago Cadena (Univ Tübingen): Deep Learning models of the rodent and primate visual system
• 1.40 - 2.15 Dylan Paiton (Univ Tübingen): Iso-response surface geometry of artificial and biological neurons [1] [2]
• 2.15 - 2.45 Poornima Ramesh (TU Munich): Adversarial Training of Neural Encoding Models on Population Spike Trains
• 2.45 - 3.15 Coffee Break
• 3.15 - 3.45 Max Burg (Univ Tübingen): Learning Divisive Normalization in Primary Visual Cortex
• 3.45 - 4.15 Dmitry Kobak (Univ Tübingen): The art of using t-SNE for single-cell transcriptomics
• 4.15 - 4.30 Open Discussion

Track IV: Robust Perception & Adversarial Robustness (Wieland Brendel, Jonas Rauber)

Venue: Max Planck House, Seminar Room

Chair: Katja Schwarz

Wieland Brendel and Jonas Rauber will give an introduction to current topics in adversarial robustness, along with a hands-on session on adversarial robustness using Foolbox. Note that participants should be comfortable with using Python and Tensorflow/Pytorch to get most out of the workshop.

Important Note for Participants of Track IV: Please bring your laptops for this workshop.