We propose novel graph pooling methods for structure-aware pooling via edge collapses.

Energy Matching unifies flow matching and energy-based models in a single time-independent scalar field, enabling efficient transport from noise to data while retaining explicit likelihood information for flexible, high-quality generation.

Current AI benchmarks suffer from systematic flaws like data leakage and selective reporting. We propose PeerBench, a community-run eval platform with secret and live tests and reputation-weighted scoring to restore trust in AI performance claims.

We present OrthoLoC, a large-scale UAV localization dataset pairing aerial imagery with orthographic geodata, enabling a new localization paradigm. Our AdHoP technique enhances matching-based localization pipelines.

We introduce EmoNet-Face, a suite of diverse, AI-generated facial emotion datasets with a novel 40-category taxonomy, to improve AI's ability to understand human emotions more accurately and empathetically.

We compare leading open SFT datasets, add quality annotations using MagPie, and design curation recipes leading to a high-performing leaner SFT mixture

We enhance diffusion models to be able to recall content from long back in a sequence in order to produce consistent content

We show the effectiveness of data-augmentation for mitigating bias due to unobserved confounding, and this motivates the proposal of our novel method for the same.

EgoExOR is the first dataset to combine egocentric and exocentric views in the OR, with gaze, hand pose, video, and dense scene graphs across 84k frames, enabling holistic, human-centric surgical scene understanding.

We propose LangHOPS, the first Multimodal Large Language Model (MLLM)-based framework for open-vocabulary object–part instance segmentation.

IPFormer introduces context-adaptive instance proposals for vision-based 3D Panoptic Scene Completion, outperforming the state-of-the-art and reducing runtime by over 14x.

A benchmark to evaluate the robustness of Earth observation foundation models