🏢 City University of Hong Kong

GLMix: A novel vision backbone efficiently integrates convolutions and multi-head self-attention at different granularities, achieving state-of-the-art performance while addressing scalability issues.

PrefPaint: Aligning image inpainting diffusion models with human preferences using reinforcement learning, resulting in significantly improved visual appeal.

MSVMamba: A novel multi-scale vision model leveraging state-space models, achieves high accuracy in image classification and object detection while maintaining linear complexity, solving the long-rang…

Boosting Robust Few-Shot Learning with Adversarial Meta-Tuning!

LuSh-NeRF: A novel model reconstructs sharp, bright NeRFs from hand-held low-light photos by sequentially modeling and removing noise and blur, outperforming existing methods.

LLM-ESR enhances sequential recommendation by integrating semantic information from LLMs, significantly improving performance on long-tail users and items.

Meta-learning a hypernetwork on CutMix-augmented data enables data-efficient and precise correction of vision transformer errors by identifying optimal parameters for fine-tuning.

G3: A novel framework leverages Retrieval-Augmented Generation to achieve highly accurate worldwide image geolocalization, overcoming limitations of existing methods.

OLIVINE uses visual foundation models for fine-grained image-to-LiDAR contrastive distillation, mitigating self-conflict issues and improving 3D representation learning.

CODA: A novel modeling scheme tackles subpopulation shifts in machine learning by disentangling spurious correlations, augmenting data strategically, and using reweighted consistency loss for improved…

PCNet boosts weakly-supervised referring image segmentation by progressively processing textual cues, mimicking human comprehension, and significantly improving target localization.

BehaviorGPT, a novel autoregressive Transformer, simulates realistic traffic agent behavior by modeling each time step as ‘current’, achieving top results in the 2024 Waymo Open Sim Agents Challenge.

Attention boosted Individualized Regression (AIR) provides a novel individualized modeling framework for matrix data, leveraging sample-specific internal relations without needing extra sample similar…

Compare2Score: A novel IQA model teaches large multimodal models to translate comparative image quality judgments into continuous quality scores, significantly outperforming existing methods.

ATAC-Diff: A versatile diffusion model for high-quality single-cell ATAC-seq data generation and analysis, surpassing state-of-the-art.