Posters

This study provides a unified theory for kernel ridge regression’s learning curve, improving existing bounds and validating the Gaussian Equivalence Property under minimal assumptions.

This paper introduces a compositional framework for algebraic circuits, deriving novel tractability conditions for compositional inference queries and unifying existing results.

A new combinatorial algorithm dramatically speeds up semi-discrete optimal transport calculations, offering an efficient solution for large datasets and higher dimensions.

Leaving the CLS token of a Vision Transformer randomly initialized during cross-domain few-shot learning consistently improves performance; a novel method leveraging this phenomenon achieves state-of-…

Debunking a common myth, this paper proves that AUPRC is not superior to AUROC for imbalanced datasets, and in fact, can worsen algorithmic bias.

Researchers unveil how causal text encoding in text-to-image models leads to information loss and bias, proposing a novel training-free optimization method that significantly improves information bala…

Canonicalization simplifies invariant and equivariant learning by connecting frames to canonical forms, leading to novel, superior frame designs for eigenvector symmetries.

Solved a long-standing open problem: Factorized ADABOOST.MH now has a proven convergence rate!

New best-of-both-worlds bandit algorithm tolerates arbitrary excessive delays, overcoming limitations of prior work that required prior knowledge of maximal delay and suffered linear regret dependence…

BADS: a novel Bayesian approach to data point selection efficiently optimizes deep learning models by jointly inferring instance weights and model parameters using stochastic gradient Langevin dynamic…

FedBNN: a novel Bayesian framework for personalized federated learning, achieves superior performance in heterogeneous settings while ensuring strict privacy via differential privacy.

4Real: Photorealistic 4D scene generation from text prompts using video diffusion models, exceeding object-centric approaches for higher realism and efficiency.

4M-21 achieves any-to-any predictions across 21 diverse vision modalities using a single model, exceeding prior state-of-the-art performance.

4Diffusion generates high-quality, temporally consistent 4D content from monocular videos using a unified multi-view diffusion model and novel loss functions.

Uncertainty-aware 4D Gaussian Splatting enhances dynamic scene reconstruction from monocular videos by selectively applying regularization to uncertain regions, improving both novel view synthesis and…

4-bit Shampoo achieves comparable performance to its 32-bit counterpart while drastically reducing memory usage, enabling efficient training of significantly larger neural networks.

3DET-Mamba: A novel end-to-end 3D object detector leveraging the Mamba state space model for efficient and accurate object detection in complex indoor scenes, outperforming previous 3DETR models.

FlowDPO: Revolutionizing 3D structure prediction with flexible probability paths & Direct Preference Optimization for enhanced accuracy and reduced hallucinations.

Learned fragment pruning accelerates 3D Gaussian splatting rendering by selectively removing fragments, achieving up to 1.71x speedup on edge GPUs and 0.16 PSNR improvement.

3DFMNet: A novel two-stage network for multi-instance point cloud registration, achieving state-of-the-art accuracy by focusing on object centers first and then performing pairwise registration.