Prototype-Based Explanations

A vision model labels a photograph “chapel.” Instead of returning a heatmap, it shows training images with similar internal representations.

These prototypes answer a familiar question: what known cases does this look like? Criticisms show nearby cases the prototype set fails to represent — instances close in feature space to a prototype that the prototype set does not explain well.

Compare Prototypes and Criticisms

Step through nearby training examples in feature space. Similarity provides the ranking; the criticism shows what the current prototype set fails to represent.

Chapel ADuke Chapel in winter, Durham NC. Representative prototype. Wikipedia

A prototype is evidence about resemblance. It does not prove the model used the same human-recognizable property.

Similarity in feature space

Prototypes are selected using distance between internal feature vectors, often Euclidean or cosine distance. MMD-Critic instead chooses a set that reduces the discrepancy between the prototype distribution and the full data distribution.

MMD-Critic — Algorithm Overview

Step 1Choose counts

Decide how many prototypes m and criticisms c you want. These are the only free parameters.

Prototypes summarize the data distribution; criticisms highlight where prototypes fall short. Choosing m and c is a design decision — more prototypes give a richer summary but are harder to inspect.

m — Prototypes

Cover the typical cases. Start with 3–10% of dataset size.

c — Criticisms

Flag edge cases. Often set to the same count as m.

1

Choose counts

Decide how many prototypes m and criticisms c you want. Thes

2

Select prototypes

Run a greedy search over the data. At each step, pick the po

3

Select criticisms

Compute the witness function for every non-prototype point.

4

Return results

The final prototype set summarizes the dataset. The final cr

MMD-Critic makes two greedy passes: one minimizing MMD² for prototypes, one maximizing the witness function for criticisms. Both passes operate over actual data points.

Examples are concrete, which helps in medicine, fraud review, and quality control. Their familiarity can also over-persuade. A close embedding does not guarantee the model matched on the property a person notices.

Show similarity scores, include contrastive cases, and check whether the selected examples cover important subgroups.

Prototypes and criticisms are always actual instances from the data

Unlike synthetic explanations, prototypes and criticisms are real training examples — nothing is generated or interpolated. This grounds the explanation in the actual data the model learned from, but it also means the explanation is limited by what exists in that dataset.

Checkpoint

Why should a prototype explanation include a criticism?