feat(research): promote & scaffold the monotonicity-survey paper (TMLR)#146
Open
davorrunje wants to merge 4 commits into
Open
feat(research): promote & scaffold the monotonicity-survey paper (TMLR)#146davorrunje wants to merge 4 commits into
davorrunje wants to merge 4 commits into
Conversation
Promotes `survey-monotonicity-ml` from the portfolio backlog to a paper root (paper-exploration). A TMLR survey of monotonicity methods in ML: taxonomy by guarantee mechanism + an original reproducible benchmark (reusing the mononet harness + datasets.yml registry) + a synthesis of universal-approximation theorems. Doubles as the thesis background / related-work foundation and the PhD qualifying-exam milestone. Scaffolds docs/research/survey-monotonicity-ml/ (paper/pitch.md seeded from the backlog row + author scope decisions; backlog.md with seed empirical-claim rows), registers the paper in docs/research/papers.md, and records the parked→ranked→promoted transition in the portfolio backlog. Next (paper-synthesis): run `literature scout --level paper` to confirm coverage/gap and seed the method inventory, then positioning → outline. Thesis wiring (background chapter + milestone) is part of #139. Generated-with: honest-scholar (https://github.com/davorrunje/honest-scholar) HonestScholar-Skill: paper-exploration Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Runs `literature scout --level paper` (keyless OpenAlex, polite pool) over 5 anchors — CMNN, Sartor 2025, Cano 2019, Daniels & Velikova 2010, Deep Lattice Networks — pulling 243 unique forward-citing works (28 multi-anchor). Classifies them by guarantee mechanism and hand-picks 14 genuine method- contribution leads, including new-since-2023 families: MonoKAN (certified monotonic KAN), MoST (monotone set transformer), MCNet (monotonic calibration), isotonic-embedding nets, and counterexample/positivity certification. Report: docs/research/survey-monotonicity-ml/paper/scout-2026-07-22.md (with full provenance + OpenAlex ids). Honestly flags the recall caveat: CMNN/Sartor/ Cano resolved to arXiv DOIs (citation split), so the completeness sweep needs published-version anchors + S2_API_KEY (for SciCite intents). Pitch blockers/ next-steps updated accordingly. Generated-with: honest-scholar (https://github.com/davorrunje/honest-scholar) HonestScholar-Skill: literature Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Completeness sweep for the monotonicity survey scout. Verified the three suspect anchors' OpenAlex records: CMNN (W4281571206) and Sartor (W4415031780) have NO separate PMLR/published record in OpenAlex — only the arXiv preprints; Cano's anchor (W2901948247) was already the published Neurocomputing record. The hypothesized arXiv-vs-published citation split therefore does not exist in OpenAlex; published-record recovery was 0 and the union corpus stays 243 unique (28 multi-anchor), identical to the first pass. The low forward-cite counts are intrinsic OpenAlex under-coverage, not a DOI split — only S2 (currently HTTP 403) or manual anchor seeding can close it. Added 7 method leads found by re-scanning the corpus (Lipschitz provably-monotonic nets W3216653328, monotone-net approximation theory W4394994853, MonoNet W4321611068, logic-based certified monotonic classifiers, input-relational verification, monotonic embeddings, partially-monotone density estimation). Co-citation/coupling neighbor probe on the sparse anchors returned no method leads (dominated by one structural-engineering application citer) — documented. New file scout-2026-07-22-sweep.md supersedes/deltas the first pass without overwriting it; pitch.md next-steps updated (sweep done; S2 contexts pending). Generated-with: honest-scholar (https://github.com/davorrunje/honest-scholar) HonestScholar-Skill: literature Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #146 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 18 18
Lines 867 867
Branches 148 148
=========================================
Hits 867 867 🚀 New features to boost your workflow:
|
Runs the deep-research workflow (107-agent web fan-out → fetch → 3-vote adversarial verification, 2/3-refute-to-kill → cited synthesis; 12 verified findings). Complements the OpenAlex citation scout with sources its forward-graph misses — especially formal-methods / PL venues (FMICS, OOPSLA). Surfaces 5 new leads the scout didn't: Igel "Smooth Min-Max Monotonic Networks" (ICML 2024), Vidot "Formal Monotony Analysis" (FMICS 2022), Monteiro "Monotonicity Regularization" (UAI 2022), Wang "Monotone Cubic B-Splines" (2023), Burdakov & Sysoev "Smoothed Monotonic Regression" (2017). Report: docs/research/survey-monotonicity-ml/paper/deep-research-2026-07-22.md (cross-checked NEW-vs-known, full verified claims + evidence + sources, caveats incl. one refuted sub-claim). Leads only — inclusion/role are the author's triage calls. Pitch next-steps updated. Generated-with: honest-scholar (https://github.com/davorrunje/honest-scholar) HonestScholar-Skill: literature Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Promotes and scaffolds a new paper —
survey-monotonicity-ml, a TMLR surveyof monotonicity methods in machine learning — via the honest-scholar
paper-explorationflow. It expands the related-work summary from the CMNN paper(arXiv:2205.11775) into a standalone survey and serves as the PhD
qualifying-exam paper plus the thesis's background / related-work foundation.
Details
The survey has three pillars (author scope decisions, 2026-07-22):
is obtained — constrained architectures · lattice · certification/verification
· soft/regularization · classical (isotonic/GAM/trees) · monotone flows.
protocol using the
benchmarks/harness and thedatasets.ymlregistry.Changes:
docs/research/survey-monotonicity-ml/paper/pitch.md(seeded) +backlog.md.docs/research/papers.md— registry row (backend: benchmarks, covers: —background) + scope bullet.
docs/research/portfolio-backlog.md— parked → ranked → promoted transition.Deliberately not here: no dashboard regeneration (the dashboard lives on the
open datasets PR; a
progressrun after both merge will pick up the new paper),and no thesis-framing edits (background-chapter + milestone wiring is tracked in
#139).
Next step (paper-synthesis):
literature scout --level paperto confirm thepost-2019 gap and seed the method inventory, then positioning → outline.
🤖 Generated with Claude Code