Research map / Learning theory and algorithms / Kernels and regression theory
Tanimoto Random Features for Scalable Molecular Machine Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use random-feature GP instead of exact GP on a random subset of M points | T_MM kernel · dockstring F2 | average test log probability: -0.951 -> -0.658 | Helped |
| Use random-feature GP instead of random subset GP | T_DP kernel · dockstring F2 | average test log probability: -0.940 -> -0.513 | Helped |
| Use RFGP (Rademacher) instead of random subset GP | T_MM kernel · dockstring KIT | average test log probability: -1.094 -> -1.222 | Hurt |
| Use data-dependent SVGP instead of data-oblivious random features | T_DP kernel · dockstring F2 | average test log probability: -0.513 -> -0.459 | Helped |
| Use SVGP instead of RFGP (Rademacher) | T_MM kernel · dockstring KIT | average test log probability: -1.222 -> -0.846 | Helped |
| Gaussian instead of Rademacher Xi in T_MM random features (ablation) | T_MM RFGP · dockstring ESR2 | average test log probability: -0.954 -> -0.956 | No clear effect |
| Gaussian instead of Rademacher Xi in T_MM features (ablation) | T_MM random features · GuacaMol 1000 molecules | MSE of Gram matrix | No clear effect |
| Normalization bias correction on T_DP random features (ablation) | T_DP RFGP · dockstring PGR | average test log probability: -1.029 -> -1.028 | No clear effect |
| Sketch bias correction on T_DP random features (ablation) | T_DP RFGP · dockstring KIT | average test log probability: -0.979 -> -0.979 | No clear effect |
| Normalization bias correction for truncated power series (ablation) | T_DP random features · GuacaMol binary fingerprints | MSE of Gram matrix | Hurt |
| Use T_DP kernel instead of T_MM kernel | RFGP · dockstring KIT | average test log probability: -1.222 -> -0.979 | Helped |
| Allocate more features to low-order terms (m_r proportional to r^-1) (ablation) | T_DP random features · GuacaMol binary fingerprints | MSE of Gram matrix | Helped |
Closest papers by meaning
- Random Features for Shift-Invariant Kernels with Moment Matching (AAAI 2017 · similarity 0.82)
- TRF: Learning Kernels with Tuned Random Features (AAAI 2022 · similarity 0.82)
- Learning Bounds for Greedy Approximation with Explicit Feature Maps from Multiple Kernels (NeurIPS 2018 · similarity 0.82)
- CheMixHub: Datasets and Benchmarks for Chemical Mixture Property Prediction (NeurIPS 2025 · similarity 0.81)
- Random features for Grassmannian kernel approximation with bounded rank-one projections (NeurIPS 2026 · similarity 0.81)
- GLACIER: Rethinking Mass Spectrum Prediction as an Object Detection Problem (NeurIPS 2026 · 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.