Posters

DGES, a novel framework, efficiently detects & handles deterministic relations in causal discovery, enhancing accuracy and scalability for real-world applications.

This paper introduces a novel notion of intrinsic similarity between language encoders, based on affine homotopy, and demonstrates its strong correlation with extrinsic similarity (downstream task per…

Improved unsupervised domain adaptation framework achieves superior performance via refined f-divergence and novel f-domain discrepancy, enabling faster algorithms and tighter generalization bounds.

OmniTokenizer: A transformer-based tokenizer achieving state-of-the-art image and video reconstruction by leveraging a novel spatial-temporal decoupled architecture and progressive training strategy.

OmniJARVIS: Unified vision-language-action tokenization enables open-world instruction-following agents via unified multimodal interaction data.

Omnigrasp: A novel RL-based method enables simulated humanoids to grasp diverse objects and precisely follow complex trajectories, advancing realistic human-object interaction in virtual environments.

OMG-LLaVA: A single model elegantly bridges image, object, and pixel-level reasoning for superior visual understanding.

Oja’s algorithm achieves minimax optimal error rates for streaming sparse PCA using a simple single-pass thresholding method, requiring only O(d) space and O(nd) time.

Offline RL agents often fail in real-world scenarios due to unseen test states. SCAS, a novel method, simultaneously corrects OOD states to high-value, in-distribution states and suppresses risky OOD …

LOLIPOP: A novel algorithm achieving near-optimal regret for offline contextual Markov Decision Processes (CMDPs) using only O(H log T) offline density estimation oracle calls.

This paper introduces Offline Behavior Distillation (OBD) to synthesize compact expert behavioral data from massive sub-optimal RL data, enabling faster policy learning.

First-Glance Off-Policy Selection (FPS) revolutionizes human-centric AI by enabling personalized policy selection for new participants without prior data, improving learning and healthcare outcomes.

This paper develops novel finite-sample bounds for off-policy linear treatment effect estimation with adaptively collected data, proposing online learning algorithms to improve estimation accuracy and…

DARAIL, a novel algorithm, tackles off-dynamics reinforcement learning by combining reward modification with imitation learning to transfer a learned policy from a source to a target domain. This app…

ODGS: Lightning-fast 3D scene reconstruction from single omnidirectional images using 3D Gaussian splatting, achieving 100x speedup over NeRF-based methods.

ODGEN: Boosting object detection accuracy by generating high-quality synthetic images using diffusion models conditioned on bounding boxes and text descriptions.

OctreeOcc uses octree queries for efficient and multi-granularity 3D occupancy prediction, surpassing state-of-the-art methods with reduced computational costs.

Octopus, a novel multi-modal LLM, uses parallel visual recognition and sequential understanding to achieve 5x speedup on visual grounding and improved accuracy on various MLLM tasks.

Model-free policy gradient methods using occupancy functions are developed for online and offline RL, achieving computational efficiency and handling arbitrary data distributions.

OccFusion: High-fidelity human rendering from videos, even with occlusions, using 3D Gaussian splatting and 2D diffusion priors.