Interpretable Representation Learning for Motion Forecasting
Buch in deiner Nähe kaufen
Beschreibung
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.
Autor*in
Royden Wagner
Themen in »Interpretable Representation Learning for Motion Forecasting«
Maschinelles Lernen Bewegungsvorhersage Autonomes Fahren representation learning mechanistic interpretability motion forecasting transformer models self-driving cars