Skip to content
Home / Catalog / Affinity-Graph Subset Audit for Spectral Representation Bounds

Affinity-Graph Subset Audit for Spectral Representation Bounds

  • Task ID: computer_science.affinity_subset_bound_audit
  • Domain: computer_science
  • Subdomain: self_supervised_learning_theory
  • Status: final
  • Benchmark set: seed42 (60 tasks)
  • Tags: contrastive_learning, self_supervised_learning, spectral_clustering, graph_diagnostics, generalization_bounds, difficult_examples

Runtime and requirements

  • Estimated time: 60-120 minutes
  • Python: >=3.10
  • Packages: numpy>=1.24, pandas>=2.0, scipy>=1.10, matplotlib>=3.7, scikit-learn==1.8.0
  • GPU required: no
  • Network required: no

Public input and output contract

Inputs

  • metadata.json (data): Agent-facing schema, item ordering, and candidate block-count range for an anonymized affinity audit.
  • affinity_matrix.npy (data): Symmetric pair-affinity matrix for unlabeled representation items.
  • screening_table.csv (data): Item ids and simple row-level affinity summaries computed from the public matrix.

Outputs

  • analysis.py (code): Agent analysis script that reads data/ and writes all required results.
  • results/group_labels.csv (data): Latent block assignment for every item, with columns item_id and group_label.
  • results/subset_scores.csv (data): Per-item disruption score and binary selected flag, with columns item_id, disruption_score, selected.
  • results/pair_regime_summary.csv (data): Affinity-regime summary table for same-block, background cross-block, and selected cross-block pairs.
  • results/bound_comparison.csv (data): Spectral-bound audit for raw, removed-subset, margin-adjusted, and temperature-scaled conditions.
  • results/spectral_report.csv (data): Eigenvalue and eigengap diagnostics for the raw and mitigated affinity graphs.
  • results/adjusted_affinity.npz (data): Adjusted normalized affinity matrices and selected index arrays used in the bound audit.
  • results/diagnostics.png (figure): Diagnostic figure summarizing the affinity matrix, spectrum, selected subset scores, and bound comparison.

Public repository files

View this task in ASI-Bench

Formal benchmark task directories in the public repository are metadata-only. Versioned prompts and inputs are distributed through the pinned benchmark dataset.

Provenance

This task is included in the current seed42 benchmark set at dataset revision f11a199f71fb4b854ac43a1bf548d5df519141a9.

This page is generated only from files tracked in the public ASI-Bench repository at commit f18382f03faf. Submission bundles, run logs, private scoring configuration, and private reference answers are not read by this page generator.