Research map / Label-efficient and robust learning / Recognition and long-tail
Probabilistic Hash Embeddings for Online Learning of Categorical Features
Aodong Li, Abishek Sankararaman, Balakrishnan Murali Narayanaswamy
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 |
|---|---|---|---|
| replace deterministic hash embeddings with probabilistic hash embeddings and Bayesian online updates | SlowAda hash embedding · Adult online classification | accuracy (x100): 82.2 -> 84.1 | Improved |
| same, vs FastAda | FastAda · Adult | accuracy (x100): 71.1 -> 84.1 | Improved |
| same on Retail sequence modeling | SlowAda · Retail | MAE: 49.1 -> 3.0 | Improved |
| same on MovieLens-32M | SlowAda · MovieLens-32M | MAE (x100): 15.3 -> 14.7 | Improved |
| PHE at fixed small memory vs collision-free expandable embeddings P-EE | P-EE · Retail | MAE: 3.2 -> 3.0 | Improved |
| PHE vs collision-free P-EE on Adult | P-EE · Adult | accuracy (x100): 84.8 -> 84.1 | Hurt |
| PHE vs EE on Bank | EE · Bank | accuracy (x100): 90.0 -> 89.6 | No clear effect |
| more training epochs for deterministic hash embeddings (FastAda) | SlowAda · Adult | accuracy (x100): 82.2 -> 71.1 | Hurt |
Closest papers by meaning
- Unified Embedding: Battle-Tested Feature Representations for Web-Scale ML Systems (NeurIPS 2023 · similarity 0.87)
- Hash Embeddings for Efficient Word Representations (NeurIPS 2017 · similarity 0.87)
- Prototypical Hash Encoding for On-the-Fly Fine-Grained Category Discovery (NeurIPS 2024 · similarity 0.86)
- NUDGE: Lightweight Non-Parametric Fine-Tuning of Embeddings for Retrieval (ICLR 2025 · similarity 0.85)
- KALAHash: Knowledge-Anchored Low-Resource Adaptation for Deep Hashing (AAAI 2025 · similarity 0.84)
- MoME: Mixture-of-Memory Embeddings for Context-Aware Sparse Lookup (NeurIPS 2026 · similarity 0.83)
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.
