148_1000.jpg -

Applying t-SNE or UMAP to see where this image sits relative to its assigned class.

(e.g., ImageNet, a local project, or a specific website?) 148_1000.jpg

Using a pre-trained ResNet-50 or Vision Transformer (ViT) to extract the embedding vector for 148_1000.jpg . Applying t-SNE or UMAP to see where this

To investigate the representational value of specific data points within the broader training set. 2. Methodology a local project

Measuring the cross-entropy loss contribution of this single image during a training epoch.

(e.g., Computer Science, Art History, or Forensics?)

(e.g., An animal, a vehicle, a medical scan?)