Your Out-of-Distribution Detection Method is Not Robust!

Readers respond to Burkhard Bilger's piece about turbulence, Gideon Lewis-Kraus's article on Anthropic, Ava Kofman's story concerning surrogacy, and Katy Waldman's essay about fawning. Burkhard Bilger's recent story about aviation turbulence opens with a dramatic account of a Singapore Airlines flight, SQ321, in May, 2024 (" Buckle Up," March 9th). The plane hit clear-air turbulence over Myanmar's Irrawaddy River, causing it to drop almost two hundred feet in an instant. During the Second World War, U.S. Army Air Forces transport planes confronted the same weather system. Flying from northeast India, over "the Hump" of intervening mountain ranges, to southwestern China, pilots routinely encountered turbulence that dropped and lifted their aircraft not hundreds of feet but thousands.

Spatio-temporal forecasting is fundamental to intelligent systems in transportation, climate science, and urban planning. However, training deep learning models on the massive, often redundant, datasets from these domains presents a significant computational bottleneck. Existing solutions typically focus on optimizing model architectures or optimizers, while overlooking the inherent inefficiency of the training data itself. This conventional approach of iterating over the entire static dataset each epoch wastes considerable resources on easy-to-learn or repetitive samples. In this paper, we explore a novel training-efficiency techniques, namely learning from complexity with dynamic sample pruning, ST-Prune, for spatio-temporal forecasting. Through dynamic sample pruning, we aim to intelligently identify the most informative samples based on the model's real-time learning state, thereby accelerating convergence and improving training efficiency. Extensive experiments conducted on real-world spatio-temporal datasets show that ST-Prune significantly accelerates the training speed while maintaining or even improving the model performance, and it also has scalability and universality.

We introduce Dataset Grouper, a library to create large-scale group-structured (e.g., federated) datasets, enabling federated learning simulation at the scale of foundation models. This library facilitates the creation of group-structured versions of existing datasets based on user-specified partitions, and directly leads to a variety of useful heterogeneous datasets that can be plugged into existing software frameworks. Dataset Grouper offers three key advantages. First, it scales to settings where even a single group's dataset is too large to fit in memory. Second, it provides flexibility, both in choosing the base (non-partitioned) dataset and in defining partitions.

Our key insight is that the user's examination and click behavior

As an extension of traditional visual single object tracking (SOT) task [2, 3, 4], VLT can harness the complementary advantages of multiple modalities. Therefore, vision-language trackers (VLTs) have the potential to achieve more promising tracking performance, which has recently attracted widespreadattention[5,6,7,8].