Lidar-assisted control (LAC) of wind turbines is a control concept that takes advantage of a nacelle-mounted lidar (a remote sensing device) to measure upstream wind speeds of a turbine to allow a preview of the incoming turbulence. Extensive ablations show that FRL improves the OOD detection performance while preserving the image generation quality.
On a new large-scale evaluation task, FRL achieves the state-of-the-art performance, outperforming a strong baseline Likelihood Regret by 10.7% (AUROC) while achieving 147$\times$ faster inference speed. FRL effectively improves performance on a wide range of generative architectures, including variational auto-encoder, GLOW, and PixelCNN++. To address the issue, we propose a novel frequency-regularized learning FRL framework for OOD detection, which incorporates high-frequency information into training and guides the model to focus on semantically relevant features. In particular, generative models are shown to overly rely on the background information to estimate the likelihood. While much research attention has been placed on defining new test-time measures of OOD uncertainty, these methods do not fundamentally change how deep generative models are regularized and optimized in training. Modern deep generative models can assign high likelihood to inputs drawn from outside the training distribution, posing threats to models in open-world deployments.
0 kommentar(er)
