imagenet-1k
ImageNet-1K refers specifically to the subset of the ImageNet dataset that contains 1,000 distinct classes, commonly used for training and evaluating image recognition models.
- Beyond Random: Automatic Inner-loop Optimization in Dataset Distillation
- Ditch the Denoiser: Emergence of Noise Robustness in Self-Supervised Learning from Data Curriculum
- Enhancing Sample Selection Against Label Noise by Cutting Mislabeled Easy Examples
- FastDINOv2: Frequency Based Curriculum Learning Improves Robustness and Training Speed
- Linear Differential Vision Transformer: Learning Visual Contrasts via Pairwise Differentials
- Quantifying Task-relevant Similarities in Representations Using Decision Variable Correlations
- REOrdering Patches Improves Vision Models
- Revisiting Residual Connections: Orthogonal Updates for Stable and Efficient Deep Networks
- Scaling Up Parameter Generation: A Recurrent Diffusion Approach
- Variational Supervised Contrastive Learning
- Vulnerable Data-Aware Adversarial Training