IPAB Workshop-15/02/2024

Speaker: Ondrej Bohdal 

Title - Feed-Forward Latent Domain Adaptation

Abstract - We study a new highly-practical problem setting that enables resource-constrained edge devices to adapt a pre-trained model to their local data distributions. Recognizing that device's data are likely to come from multiple latent domains that include a mixture of unlabelled domain-relevant and domain-irrelevant examples, we focus on the comparatively under-studied problem of latent domain adaptation. Considering limitations of edge devices, we aim to only use a pre-trained model and adapt it in a feed-forward way, without using back-propagation and without access to the source data. Modelling these realistic constraints bring us to the novel and practically important problem setting of feed-forward latent domain adaptation. Our solution is to meta-learn a network capable of embedding the mixed-relevance target dataset and dynamically adapting inference for target examples using cross-attention. The resulting framework leads to consistent improvements over strong ERM baselines. We also show that our framework sometimes even improves on the upper bound of domain-supervised adaptation, where only domain-relevant instances are provided for adaptation. This suggests that human annotated domain labels may not always be optimal, and raises the possibility of doing better through automated instance selection.

Speaker - Yuhui Lin

Title - Topological scenario analysis for autonomous vehicle, review and extension

Abstract - To ensure and enhance functional safety in autonomous vehicle development, an Operational Design Domain (ODD)-based process has been proposed. This scenario-driven design and testing approach, utilising machine-readable scenario descriptions e.g. OPENSCENARIO,. However, existing scenarios are often categorised at a high level, focusing on features such as the ego car's operation or the number of obstacles. This talk reviews current topological scenario analysis techniques that offer abstract yet intuitive geometric representations and classifications of trajectories and ego behaviours in scenarios. The discussion also considers potential extensions of these techniques to analyse dynamic objects in autonomous vehicle scenarios.