Conflict-Driven Learning in AI Planning State-Space Search - Code and Benchmarks
Bibliographic record
Abstract
This archive encompasses the source code, the benchmarks, and the experiment scripts, accompanying our work on conflict-driven learning in state-space search in classical and probabilistic planning. The source code is based on the Fast Downward (FD) classical planning system https://fast-downward.org/, and features extensions to support MDP goal-probability analysis; state-space search methods (both for the deterministic and the probabilistic case) to identify conflicts, i.e., dead-end states, during the search for a solution; and various dead-end detection (mostly based on classical planning heuristics) and refinement algorithms, used for pruning recognized dead-end states during search, respectively for improving the dead-end detector by learning from the identified conflicts. The benchmarks are composed of A PPDDL benchmark collection, which aggregates known IPPC benchmarks (with minor modifications addressing unsupported features); newly generated network penetration testing benchmarks; as well as Canadian (adding ``road-graph uncertainty'') variants of well-known classical-planning resource-constrained benchmarks (NoMystery, Rovers, TPP). A collection of solvable classical planning benchmarks containing dead-end states, aggregating known IPC and resource-constraint benchmarks. A collection of unsolvable classical planning benchmarks, comprising the UIPC benchmarks and unsolvable variants of the resource-constrained benchmarks from the solvable part. The experiment scripts should allow reproducing our results. (Some external planners to which we compare are however not included, and need to be downloaded separately.)
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.014 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.033 | 0.009 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".