Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Gaussian Process-Gated Hierarchical Mixtures of Experts
Yuhao Liu, Marzieh Ajirak, Petar M. Djurić
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 |
|---|---|---|---|
| use OF1 linear-pooling objective instead of normalized likelihood OF2 (ablation) | GPHME · MNIST | accuracy (%): 97.79 -> 98.49 | Improved |
| replace BHME with GPHME (GP gating via random features) | tree-based HME · RIN | accuracy (%): 76.71 -> 98.05 | Improved |
| replace BHME with GPHME | tree-based HME · MAG | accuracy (%): 79.25 -> 87.83 | Improved |
| replace BHME with GPHME | tree-based HME · PEN | accuracy (%): 95.53 -> 99.54 | Improved |
| replace BHME with GPHME | tree-based HME · CON | MSE: 0.147 -> 0.098 | Improved |
| replace BHME with GPHME | tree-based HME · ABA | training time (min): 42.9456 -> 10.7126 | Improved |
| use GPHME instead of BHME | tree-based HME · TWO | accuracy (%): 98.06 -> 98.01 | No clear effect |
| use NIS-N per-node feature spaces rather than shared ISO-N (ablation) | GPHME · Optical Digits and Protein | RMSE / error rate | No clear effect |
| share feature space per level (ISO-L) rather than per node or global (ablation) | GPHME · Optical Digits and Protein | RMSE / error rate | Hurt |
| increase tree height (ablation) | GPHME · Optical Digits and Protein | RMSE / error rate | Mixed |
| use ARC-COSINE kernel instead of RBF (ablation) | GPHME · UCI tasks | accuracy | Mixed |
| increase tree height from 1 to 4 (ablation) | GPHME · AIRLINE | accuracy: 71.79 -> 72 | Improved |
Closest papers by meaning
- Mixtures of Gaussian Process Experts with SMC^2 (JMLR 2025 · similarity 0.88)
- Healing Products of Gaussian Process Experts (ICML 2020 · similarity 0.87)
- Convergence Rates for Gaussian Mixtures of Experts (JMLR 2022 · similarity 0.87)
- Sigmoid Gating is More Sample Efficient than Softmax Gating in Mixture of Experts (NeurIPS 2024 · similarity 0.85)
- Statistical Perspective of Top-K Sparse Softmax Gating Mixture of Experts (ICLR 2024 · similarity 0.84)
- A General Theory for Softmax Gating Multinomial Logistic Mixture of Experts (ICML 2024 · similarity 0.84)
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.
