Informed Spectral Normalized Gaussian Processes for Trajectory Prediction
This work addresses the problem of computational inefficiency in informed learning for autonomous driving, offering an incremental improvement over existing SNGP methods.
The paper tackles the computational expense of using informative priors in probabilistic deep learning by proposing a regularization-based continual learning method for spectral normalized Gaussian processes (SNGPs), enabling integration of prior knowledge without rehearsal memory. It demonstrates increased data-efficiency and robustness in trajectory prediction for autonomous driving on public datasets.
Prior parameter distributions provide an elegant way to represent prior expert and world knowledge for informed learning. Previous work has shown that using such informative priors to regularize probabilistic deep learning (DL) models increases their performance and data-efficiency. However, commonly used sampling-based approximations for probabilistic DL models can be computationally expensive, requiring multiple inference passes and longer training times. Promising alternatives are compute-efficient last layer kernel approximations like spectral normalized Gaussian processes (SNGPs). We propose a novel regularization-based continual learning method for SNGPs, which enables the use of informative priors that represent prior knowledge learned from previous tasks. Our proposal builds upon well-established methods and requires no rehearsal memory or parameter expansion. We apply our informed SNGP model to the trajectory prediction problem in autonomous driving by integrating prior drivability knowledge. On two public datasets, we investigate its performance under diminishing training data and across locations, and thereby demonstrate an increase in data-efficiency and robustness to location-transfers over non-informed and informed baselines.