An ice-fishing shack tags each hole photo “safe,” “thin,” or “open water.” How should those categorical tags be made neural-network ready for a training loss?
Select an answer to reveal the explanation.
Short Explanation
Losses count numbers, not sticky-note words. Turn “safe / thin / open water” into a one-hot or a small embedding—do not shove the phrase into the pixel tensor.
Full Explanation
Categorical labels must be numeric for a supervised loss—commonly one-hot vectors or learned label embeddings. Embedding free text inside the vision tensor, omitting labels, or using unrelated metadata fails that readiness requirement.