Research map / Learning theory and algorithms / Matrix and tensor methods
Kronecker Determinantal Point Processes
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 Kronecker-factored subkernels and alternating KrK-Picard updates instead of full-kernel Picard | DPP kernel learning · GENES | Average runtime: 161.5 ± 17.7 s -> 8.9 ± 0.2 s | Helped |
| Use stochastic updates (ablation) | KrK-Picard · GENES | Average runtime: 8.9 ± 0.2 s -> 1.2 ± 0.02 s | Helped |
| Use stochastic updates (ablation) | KrK-Picard · GENES | NLL increase (1st iter.): $(2.96 \pm 0.02) \cdot 10^4$ -> $(3.13 \pm 0.04) \cdot 10^4$ | Helped |
| Use KrK-Picard instead of Picard | DPP kernel learning · GENES | NLL increase (1st iter.): $(2.81 \pm 0.03) \cdot 10^4$ -> $(2.96 \pm 0.02) \cdot 10^4$ | Helped |
| Constrain the kernel to a Kronecker product | DPP kernel learning · Baby registries, train and test categories | log-likelihood: For example, apparel train: -10.2 -> -10.7; diaper test: -10.7 -> -11.2 | Hurt |
| Use alternating KrK-Picard updates instead of joint Picard-style updates (ablation) | Kronecker-factored DPP kernel learning · Synthetic tests | convergence speed | Helped |
| Use joint Picard-style updates and project back to Kronecker structure (ablation) | Kronecker-factored DPP kernel learning · Synthetic tests | convergence speed | Hurt |
Closest papers by meaning
- Sampling from a k-DPP without looking at all items (NeurIPS 2020 · similarity 0.93)
- Scalable Sampling for Nonsymmetric Determinantal Point Processes (ICLR 2022 · similarity 0.92)
- Learning Determinantal Point Processes with Moments and Cycles (ICML 2017 · similarity 0.92)
- Scalable Learning and MAP Inference for Nonsymmetric Determinantal Point Processes (ICLR 2021 · similarity 0.92)
- State-of-art minibatches via novel DPP kernels: discretization, wavelets, and rough objectives (NeurIPS 2026 · similarity 0.90)
- Small coresets via negative dependence: DPPs, linear statistics, and concentration (NeurIPS 2024 · similarity 0.88)
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.