Poly-view contrastive learning maximizes information by increasing view multiplicity, leading to improved representation learning.
MIM4D proposes a novel pre-training paradigm based on dual masked image modeling (MIM) for autonomous driving representation learning, achieving state-of-the-art performance on the nuScenes dataset.
Learning high-quality data representations from randomness is a feasible and plausible alternative when transformation invariance is challenging.
DecisionNCE는 결정을 위한 다중 모달 표현을 내재적 선호 학습을 통해 효과적으로 추출하는 통합된 표현 학습 프레임워크를 제안합니다.
Knowledge-Link Graph from LLMs improves representation learning in MTS data.
This paper provides theoretical guarantees for learning nonlinear representations from multiple tasks, even when the data distributions across tasks are different and the data within each task is statistically dependent.