- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper introduces ConceptDiff, a method for learning a small set of concept detectors that explain how a target dataset A differs from a reference dataset B. Built on a frozen pretrained encoder, the method aggregates local token evidence using learnable thresholds and trains the detectors with a target-oriented multiple-instance learning (MIL) classification objective, together with diversity, balance, and commitment regularization to encourage specialization. The authors evaluate concept recovery, discriminability, coverage, and interpretability across image, text, and video settings, as well as error slice discovery.
Critical strengths
- A practically relevant problem with a comprehensive evaluation framework Explaining dataset differences through multiple interpretable concepts has practical value for distribution shift analysis, model error diagnosis, and the analysis of generated content. The paper considers discriminability, concept coverage, semantic coherence, and stability together. This evaluation framework can expose failures such as accurate classification accompanied by concept collapse, making it more informative for assessing explanations.
- A coherent methodological design that provides localized evidence Local token selection, target-oriented MIL aggregation, and concept specialization regularizers are well aligned with the objective of jointly explaining how A differs from B using a small number of concepts. The learned detectors can score new examples and localize the evidence supporting each concept, facilitating inspection and interpretation of the results.
- Substantial improvements on selected key benchmarks On the constructed image tasks, ConceptDiff achieves an ARI of 0.460, compared with 0.275 for the strongest reported baseline. On the evaluated subset of the error slice discovery benchmark, it achieves an AP of 0.807, compared with 0.742 for the strongest baseline. These results provide empirical support for improved recovery of semantic groups, while the text and video examples demonstrate potential applicability across modalities. These findings are reported in Tables 1 and 4 of the manuscript.
Critical weaknesses
- The novelty relative to the closest prior work is not sufficiently delineated ProtoSumm [1] already uses interpretable classifiers to learn dataset differences and addresses specialization and collapse among multiple prototypes. NLParam [2] already optimizes multiple interpretable predicates under a task objective. The main methodological contribution of ConceptDiff therefore appears to lie in its particular combination of architecture and regularization. However, the manuscript does not sufficiently establish what additional capabilities this combination provides beyond the closest prior approaches, and it neither cites nor evaluates NLParam.
- Substantial baseline adaptations may affect the comparative conclusions The manuscript replaces the joint sparse feature selection in the original HypotheSAEs [3] with univariate AUROC ranking. It also replaces the semantic verifier in Zhong et al. [4]—whose adapted baseline is referred to as TextDiff in the manuscript—with text embedding similarity. These changes may affect concept complementarity and semantic scoring, making it difficult to attribute the observed performance gaps entirely to the proposed method. Their effects need to be quantified. The available evidence does not establish that the adaptations necessarily weaken the baselines.
- The independent contributions of the core architectural components are insufficiently validated Although the paper includes ablations of the regularization terms, it does not separately evaluate the contributions of the asymmetric concept architecture, MIL aggregation, and learnable threshold pooling. Moreover, some ablated configurations in Figure 2 obtain slightly higher ARI values than the full model. The current results therefore suggest tradeoffs among objectives, without establishing that every proposed component makes an independent and necessary contribution.
- Reliable concept recovery under more complex difference structures remains insufficiently established The paper includes real-world case studies and explorations with different values of K, but its strongest evidence for semantic recovery still comes from constructed tasks with known concept counts. Systematic validation remains limited when the true number of concepts is unknown, factors overlap, or concept frequencies are substantially imbalanced. In particular, whether balance and commitment regularization can cause undesirable splitting or merging of concepts under these conditions has not been sufficiently examined.
References [1] ProtoSumm Babbar, V., Guo, Z., & Rudin, C. (2025). “What is Different Between These Datasets?” A Framework for Explaining Data Distribution Shifts. Journal of Machine Learning Research, 26(180), 1–64. [2] NLParam Zhong, R., Wang, H., Klein, D., & Steinhardt, J. (2024). Explaining Datasets in Words: Statistical Models with Natural Language Parameters. Advances in Neural Information Processing Systems, 37 (NeurIPS 2024). [3] HypotheSAEs Movva, R., Peng, K., Garg, N., Kleinberg, J., & Pierson, E. (2025). Sparse Autoencoders for Hypothesis Generation. In Proceedings of the 42nd International Conference on Machine Learning, Proceedings of Machine Learning Research, 267, 44997–45023. [4] TextDiff: Original work underlying the adapted baseline Zhong, R., Snell, C., Klein, D., & Steinhardt, J. (2022). Describing Differences between Text Distributions with Natural Language. In Proceedings of the 39th International Conference on Machine Learning, Proceedings of Machine Learning Research, 162, 27099–27116.
Questions for authors
- Novelty: What specific limitation of ProtoSumm and NLParam does ConceptDiff’s main methodological innovation address?
- Baseline adaptations: Does ConceptDiff retain its main performance advantages when the baselines preserve their original core mechanisms and are evaluated under matched tuning and evaluation budgets?
- Component contributions: How much does concept recovery performance change when the core architectural components are replaced one at a time, while keeping the encoder, regularization, and training budget fixed?
- Complex difference structures: Can the method reliably recover the main differences when the true number of concepts is not provided, concepts overlap, and their frequencies are substantially imbalanced?
4 comments
W2 & Q2: Original vs. Adapted Mechanisms in HypotheSAEs and TextDiff
We would like to clarify that the adaptations made in the paper were primarily so we could easily apply the baselines across different modalities, and were never meant to affect the comparative conclusions. In fact, some of our changes were meant to be an enhancement to the baseline, such as upgrading TextDiff to use a stronger, more modern language model and embedding scorer.
That said, to answer the reviewer's query, we ran the experiments with the original baseline mechanisms restored (using the original L1 sparse feature selection for HypotheSAEs and the released semantic verifier for TextDiff). As shown in the tables below for Images (Table 1 benchmark) and Text (EditLens Gemini-generated vs. human writing, Table 2), using the original mechanisms leads to mixed trade-offs across metrics, but ConceptDiff still maintains a clear advantage overall.
Table 1: Image Benchmark (12 tasks, K = 3)
| Method | ARI ↑ | cEER ↓ | Act ↑ | Cov ↑ | ΔCoh ↑ | Stab ↑ |
|---|---|---|---|---|---|---|
| HypotheSAEs (pre.) (Paper) | 0.1172 | 0.1551 | 1.0000 | 0.5335 | 0.1176$^{\mathcal{R}}$ | 0.1573 |
| HypotheSAEs (cls.) (Paper) | 0.1310 | 0.0992 | 0.8611 | 0.7511 | 0.0651 | 0.2130 |
| HypotheSAEs (pre.) (Original) | 0.1409 | 0.3380 | 1.0000 | 0.3111 | 0.0877$^{\mathcal{R}}$ | 0.2931 |
| HypotheSAEs (cls.) (Original) | 0.0941 | 0.3910 | 0.9167 | 0.1859 | 0.0397 | 0.2757 |
| ConceptDiff (ours) | 0.4596 | 0.0617 | 1.0000 | 0.8272 | 0.1319$^{\mathcal{R}}$ | 0.5221 |
| indicates monosemanticity calculated based on the concept region. |
Table 2: Text Benchmark (EditLens: Gemini-Generated vs. Human Writing, K = 4)
| Method | cEER ↓ | Act ↑ | Cov ↑ | ΔCoh ↑ | Stab ↑ |
|---|---|---|---|---|---|
| HypotheSAEs (pre.) (Paper) | 0.0962 | 1.0000 | 0.4912 | 0.2768 | 0.0886 |
| HypotheSAEs (cls.) (Paper) | 0.0038 | 0.4167 | 0.9986 | 0.1233 | 0.0040 |
| TextDiff (Paper) | 0.2532 | 1.0000 | 0.2791 | 0.2527 | 0.1502 |
| HypotheSAEs (pre.) (Original) | 0.1329 | 0.6667 | 0.2093 | 0.2478 | -0.0230 |
| HypotheSAEs (cls.) (Original) | 0.0003 | 0.3333 | 0.6511 | 0.0928 | 0.3333 |
| TextDiff (Original) | 0.3043 | 0.7500 | 0.7974 | 0.0393 | -0.0159 |
| ConceptDiff (ours) | 0.0021 | 1.0000 | 1.0000 | 0.2265 | 0.9267 |
We hope these results confirm that ConceptDiff's improvements hold regardless of whether the adapted or original baseline mechanisms are used, and we will include this comparison in the revised paper.
W3 & Q3: Ablations of Core Architectural Components (Asymmetric Head, MIL Pooling, Threshold Routing)
Following the reviewer's suggestion, we ran an architectural ablation study across all 12 tasks on the image benchmark (K = 3, 3 seeds) where we replace each of the core architectural components one at a time while keeping the frozen DINOv2-L encoder, regularization weights, and training budget fixed.
First, we tested a symmetric target and reference head that replaces our asymmetric target-only head and learned detectors for both dataset A and reference dataset B. Second, we tested a linear classifier over concepts that replaces our MIL smooth-max aggregation with a standard linear classifier over the concept scores. Third, we tested mean pooling over tokens that replaces our thresholded token routing with average pooling across all tokens. And as shown in the table below, replacing any of the three components hurts group recovery (ARI), separation (cEER), coverage (Cov), or stability (Stab).
| Architecture Variant | ARI ↑ | cEER ↓ | Act ↑ | Cov ↑ | ΔCoh ↑ | Stab ↑ |
|---|---|---|---|---|---|---|
| Symmetric target/reference head | 0.4100 | 0.1132 | 1.0000 | 0.5672 | 0.1237 | 0.4317 |
| Linear classifier over concepts (w/o MIL) | 0.4241 | 0.0816 | 1.0000 | 0.6764 | 0.1056 | 0.4358 |
| Mean pooling over tokens (w/o threshold routing) | 0.3823 | 0.0665 | 1.0000 | 0.7889 | 0.1050 | 0.4423 |
| ConceptDiff (Full Model) | 0.4596 | 0.0617 | 1.0000 | 0.8272 | 0.1319 | 0.5221 |
Additionally, we believe there is a slight misinterpretation regarding results shown in Figure 2 of the paper. While removing both the commitment and diversity regularizers do increase ARI slightly from 0.460 to 0.472, the other evaluation aspects degrade, as separation worsens (cEER increases from 0.062 to 0.071), target coverage drops (from 0.823 to 0.808), and stability across seeds drops significantly (from 0.522 to 0.471). We want to clarify that, contrastive concept discovery is a multi-objective task where discovered differences must be discriminative, diverse, and monosemantic at the same time. Thus, looking at all metrics together is essential when evaluating a method in this task, and our full model provides the strongest overall balance across every criterion rather than over-optimizing ARI alone.
These results show that each architectural component and regularizer in ConceptDiff makes a necessary contribution to the final model. We hope this clarifies the reviewer's concern, and we will include this architectural ablation table and discussion in the final revised paper.
W4 & Q4: Discovering Dataset Differences In Real-world Scenario.
We agree with the reviewer that discovering differences in real-world scenarios is challenging, especially when the true number of concepts is unknown, factors overlap, and class frequencies can be imbalanced. This is precisely why we included our real-world case studies on EditLens text and dashcam car crash videos, where no ground-truth concept counts or balanced distributions exist.
Firstly, on how ConceptDiff handles unknown , we view as a human-defined budget that controls the resolution of the discovered differences. As we increase , the model naturally decomposes broad dataset differences into finer, more specific sub-concepts rather than collapsing or failing to converge, which is clearly visible in our text experiments where different values yield coherent sub-categories.
Additionally, as we noted in our response to Reviewer ANji, we show that stability across random seeds and cross-validation splits provides a principled way to guide the choice of in practice. When exceeds the true structural complexity of the underlying data distribution, the discovered concepts begin to split arbitrarily, leading to a sharp drop in stability across seeds. Practitioners can thus monitor stability scores across increasing values of to identify the point before instability sets in, serving as a reliable indicator for the ideal concept budget.
Thus, we believe ConceptDiff provides a robust and reliable way to uncover structure in complex real-world environments. We thank the reviewer for bringing up this important point, and we will expand our discussion in the revised paper.
References
[1] NLParam Zhong, R., Wang, H., Klein, D., & Steinhardt, J. (2024). Explaining Datasets in Words: Statistical Models with Natural Language Parameters. Advances in Neural Information Processing Systems, 37 (NeurIPS 2024).
[2] TextDiff: Original work underlying the adapted baseline Zhong, R., Snell, C., Klein, D., & Steinhardt, J. (2022). Describing Differences between Text Distributions with Natural Language. In Proceedings of the 39th International Conference on Machine Learning, Proceedings of Machine Learning Research, 162, 27099–27116.
Thank you for the additional experiments and clarifications. Restoring the original baseline mechanisms and providing controlled architectural ablations substantially alleviate my concerns about baseline adaptations and component contributions. I consider the architectural ablation concern largely addressed.
Presenting examples in descending activation order may provide cues that inflate simulation correlations, and no randomized-order results are provided to exclude this possibility. This leaves an important concern about the evidence supporting monosemanticity.
The discussion of NLParam also remains incomplete: its task-driven optimization of multiple interpretable predicates is directly relevant to the novelty claim. The architectural differences are now clearer, but the contribution should be more carefully delineated against this prior work.
The rebuttal strengthens the empirical case, and I have updated my assessment positively in that respect. Nevertheless, the remaining concerns about novelty prevent me from moving to acceptance. I therefore retain my weak-reject rating.
Thank you for the insightful review and for recognizing the practical value of our problem formulation, our multi-metric evaluation framework, and our coherent design for localizing token evidence. We appreciate your constructive suggestions regarding comparisons to closest prior work (ProtoSumm and NLParam), baseline adaptations, architectural component ablations, and robustness under imbalanced or overlapping concepts. We have answered all the questions and supplemented this rebuttal with new experiments and ablations below to address your concerns.
W1 & Q1: Delineating Novelty Relative to ProtoSumm and NLParam
Comparison with ProtoSumm. While we include ProtoSumm as a baseline across Tables 1, 3, and 4, there are two main differences in how ConceptDiff works. First, ProtoSumm is symmetric and tries to learn prototypes for both dataset A and dataset B. When dataset B is just normal background data (such as normal driving or error-free predictions), modeling B is unnecessary and hurts performance on finding what makes A different. Second, ProtoSumm forces each concept to match a single training patch. This works for simple local textures, but struggles when a concept is more abstract or spans across multiple tokens. In contrast, ConceptDiff only models what triggers dataset A (using an asymmetric MIL objective) and uses learnable thresholds to combine multiple relevant tokens per input (Eqs. 1 and 2). As shown in our results, this improves ground-truth group recovery (ARI) from 0.139 for ProtoSumm to 0.460 for ConceptDiff in Table 1, improves error slice discovery AP from 0.742 to 0.807 in Table 4, and avoids the concept collapse ProtoSumm has on crash videos (Act = 0.22 vs. our 1.00 in Table 3).
Comparison with NLParam. Thank you for pointing out NLParam [1]. It is an interesting work that extends the paradigm of TextDiff [2] (which we already evaluate in Tables 2 and 5) that fits a statistical model over natural-language predicates using an LLM. However, since NLParam is very similar in setup to TextDiff, it shares the same trade-offs for our task. It relies on querying an LLM proposer and verifier across samples during optimization, which makes it expensive to scale to large image or video datasets and cannot directly localize differences to specific visual patches or video tokens. In contrast, ConceptDiff learns continuous concept detectors directly from tokens in a few minutes without an LLM in the loop, making it complementary to text-predicate methods like NLParam. We will cite and discuss NLParam alongside TextDiff in the revised paper.
We hope this clarifies how ConceptDiff differs from both ProtoSumm and NLParam, and we will include this discussion in the revised paper.