NegCosIC: Negative Cosine Similarity-Invariance- Covariance Regularization for Few-Shot Learning
Few-shot learning continues to pose a challenge as it is inherently Accessories difficult for visual recognition models to generalize with limited labeled examples.When the training data is limited, the process of training and fine-tuning the model will be unstable and inefficient due to overfitting.In this paper, we introduce NegCosIC: Negative Co