Explore
Interpretable Representation Learning for Motion Forecasting
Royden Wagner
2026
0 Ungluers have
Faved this Work
Login to Fave
We address interpretable representation learning for motion forecasting in self-driving cars. Rather than treating transformers as black boxes, we develop methods to interpret and modify learned representations. We introduce self-supervised pre-training with interpretable objectives. Moreover, we probe latent spaces of forecasting models and reveal interpretable features, allowing us to make targeted interventions. Finally, we uncover retrocausal mechanisms, which enable goal-based instructions.
This book is included in DOAB.
Why read this book? Have your say.
You must be logged in to comment.
Rights Information
Are you the author or publisher of this work? If so, you can claim it as yours by registering as an Unglue.it rights holder.Downloads
This work has been downloaded 0 times via unglue.it ebook links.
- 0 - pdf (CC BY) at OAPEN Library.
Keywords
- autonomes Fahren
- Bewegungsvorhersage
- Maschinelles Lernen
- Mechanistic interpretability
- Motion forecasting
- Representation Learning
- Self-driving cars
- thema EDItEUR::U Computing and Information Technology::UY Computer science
- transformer models
Links
DOI: 10.5445/KSP/1000191275Editions