Research map / Learning theory and algorithms / Matrix and tensor methods
Exploiting Field Dependencies for Learning on Categorical Data
Zhibin Li, Piotr Koniusz, Lu Zhang, Dan Pagendam, Peyman Moghadam
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add meta-dependency learning (per-instance dependency refinement) (ablation) | MDL backbone · Avazu | Logloss/AUC | Improved |
| use meta-learned instance-wise dependencies instead of a global dependency regulariser (ablation) | MDL · Avazu | Logloss | Improved |
| global dependency loss as regulariser (global dep) instead of none (ablation) | MDL backbone · Avazu | Logloss | Mixed |
| add dependency learning with few fields and high embedding dimension (ablation) | MDL · ML-Tag | Logloss | No clear effect |
| increase lambda toward uniform field weights (ablation) | MDL · Avazu (val/test) | Logloss | Mixed |
| increase refinement steps T beyond 4 (ablation) | MDL · Avazu (val/test) | Logloss | Hurt |
| use MDL instead of FFM | FFM · ML-Tag | Logloss: 0.2055 -> 0.1843 | Improved |
| use MDL instead of RaFM | RaFM · Frappe | Logloss: 0.1199 -> 0.1219 | No clear effect |
| use MDL instead of FWL | FWL · Criteo | AUC: 0.8130 -> 0.8145 | Improved |
| use MDL instead of FM | FM · ML-Rating | MSE: 0.7406 -> 0.7263 | Improved |
| use tabular deep models instead of sparse-categorical models | TabNet/NODE · KDD2012 | AUC: 0.8073 -> 0.7619 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- RaFM — RaFM: Rank-Aware Factorization Machines (ICML 2019) · better on 1 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Field-wise Learning for Multi-field Categorical Data (NeurIPS 2020 · similarity 0.89)
- Neural Dependencies Emerging From Learning Massive Categories (CVPR 2023 · similarity 0.84)
- Statistical Estimation from Dependent Data (ICML 2021 · similarity 0.84)
- Learning Structured Latent Factors from Dependent Data:A Generative Model Framework from Information-Theoretic Perspective (ICML 2020 · similarity 0.83)
- The Potts-Ising model for discrete multivariate data (NeurIPS 2020 · similarity 0.83)
- Generalized Category Discovery via Reciprocal Learning and Class-Wise Distribution Regularization (ICML 2025 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
