Writing study: Soft Local Completeness: Rethinking Completeness in XAI

ICCV overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← ICCV 2025 oral writing corpus

Ziv Weiss Haddad, Oren Barkan, Yehonatan Elisha, Noam Koenigstein · Official ICCV 2025 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Training, objectives & optimization (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

11 physical pages

Before references / appendix

≈ 8.05 page-equivalents

Main figures / tables

5 / 6

Attached appendix figures / tables

0 / 0

Attached appendix span

No boundary detected; not assumed zero

Before references / appendix: ≈ 73.18%References / intervening material: ≈ 26.82%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	1	≈ 1.01 pp.	9.18%
2. Related Work	2	≈ 0.6 pp.	5.45%
3. Soft Local Completeness	2	≈ 3.43 pp.	31.18%
4. Experiments and Results	6	≈ 2.52 pp.	22.91%
5. Conclusion	8	≈ 0.2 pp.	1.82%
References	9	≈ 2.95 pp.	26.82%

All detected figure and table locations

Main text

• Figure 1 · p. 3
• Figure 2 · p. 5
• Figure 3 · p. 6
• Table 1 · p. 7
• Table 2 · p. 7
• Figure 4 · p. 8
• Figure 5 · p. 8
• Table 3 · p. 8
• Table 4 · p. 8
• Table 5 · p. 8
• Table 6 · p. 8

Attached appendix

No recognized caption labels.

Argument and organization

1. 1 Introduction; 2 Related Work

   Challenges global completeness as an insufficient explanation criterion and motivates a local soft alternative.

2. 3 Soft Local Completeness

   Uses a deliberately simplified masked-image example, then derives a regularized completeness-gap optimization.

3. 4 Experiments and Results; 5 Conclusion

   Triangulates perturbation faithfulness, controlled part interventions and human segmentation alignment across model families.

Problem, claim and contribution

A flat map can satisfy normalized global completeness, providing a concrete counterexample before proposing the new criterion.

Section 3, physical p.3 · Main paper

Argument and section order

The toy assumptions are explicitly called unrealistic before the method relaxes exact local equality into a gap objective.

Sections 3.1–3.2, pp.3–4 · Main paper

Figures, captions and visual explanation

The pipeline separates cached masked-input model responses from optimization of the attribution map, clarifying why gradient descent does not require model gradients.

Figure 2, p.5 · Main paper

Tables, comparisons and evidence

The evaluation distinguishes plain SLOC, per-model mask calibration and SLOCm, which selects an attribution using the same input’s IDD metric; segmentation non-wins remain visible.

Section 4.1, p.7; Table 6, p.8 · Main paper

Appendix and reproducibility

Appendices are assigned concrete roles: optimization details D, ablations F, complexity G, runtime H, sanity checks J and limitations M; none were inspected here.

Sections 3.2,4.2,5, pp.6–8 · Main paper

Clarity and writing conventions

The paper explicitly rejects both guaranteed exact-completeness convergence and the inference that better segmentation necessarily means a better explanation.

Section 3.2, p.6; Section 4.2, p.8 · Main paper

What to reuse—and what not to infer

Transferable lesson: Introduce an intuitive axiom with a counterexample, state where the toy argument stops applying, and triangulate its practical value with different evaluation notions.

Boundary: Soft local completeness and attractive heatmaps are not general causal-faithfulness guarantees. Perturbation metrics inherit masking assumptions, and metric-selected SLOCm is not an independent test of the selection metric. Runtime and axiomatic claims assigned to the uninspected supplement remain outside this scan.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-10-07

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Writing-oriented scan across the main manuscript, physical PDF pp.1–8, covering framing, section progression, method explanations, captions, comparison protocols and concluding scope; reference boundary checked. Selected full-page rasters were inspected. Not exhaustive line-by-line reading, proof verification or experimental reproduction.

Appendix / supplement

Main-paper appendix/supplement references were checked. A separate supplementary source is linked in the inventory but was not retrieved or inspected; its contents are not included in the reading or visual-page counts.

Visually inspected PDF pages

5, 8

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Official CVF open-access ICCV 2025 accepted-paper PDF, retrieved 7 October 2026; CVF states identical to accepted version except watermark; not assumed identical to initial submission · 11 pages