Learning the Parameters of Determinantal Point Process Kernels
Authors: Raja Hafiz Affandi, Emily Fox, Ryan Adams, Ben Taskar
ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the utility of our DPP learning methods in studying the progression of diabetic neuropathy based on the spatial distribution of nerve fibers, and in studying human perception of diversity in images. [...] 5. Experiments [...] 5.1. Simulations [...] We test our Bayesian learning algorithms on simulated data [...] 5.2. Applications |
| Researcher Affiliation | Academia | Raja Hafiz Affandi RAJARA@WHARTON.UPENN.EDU Department of Statistics, University of Pennsylvania Emily B. Fox EBFOX@STAT.WASHINGTON.EDU Department of Statistics, University of Washington Ryan P. Adams RPA@SEAS.HARVARD.EDU Department of Statistics, Harvard Univerity Ben Taskar TASKAR@CS.WASHINGTON.EDU Department of Computer Science & Engineering, University of Washington |
| Pseudocode | Yes | We highlight two techniques: random-walk Metropolis-Hastings (MH) and slice sampling. [...] See Alg. 1 of the Supplement. [...] See Alg. 2 of the Supplement. [...] We show this procedure in Alg. 3 of the Supplement. |
| Open Source Code | No | The paper does not provide any explicit statements or links indicating that the source code for the described methodology is publicly available. |
| Open Datasets | No | The paper uses simulated data and refers to datasets from previous studies or standard feature sets (Waller et al., 2011; Kulesza & Taskar, 2011a; Lowe, 1999; Vedaldi & Fulkerson, 2010; Oliva & Torralba, 2006) but does not provide concrete access information, such as a direct link, DOI, or repository for a publicly available dataset. |
| Dataset Splits | No | The paper does not provide specific details on dataset splits (e.g., exact percentages or sample counts for training, validation, and testing sets) or explicit cross-validation setup details beyond mentioning a 'leave-one-out classification' strategy in one application. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU or CPU models, memory, or cloud computing instance types used for running the experiments. |
| Software Dependencies | No | The paper mentions general software components like 'Python', 'PyTorch', 'CUDA', 'scikit-learn', 'VLFeat', but does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | Yes | The hyperparameters of f(ˆΘ|Θi) tune the width of the distribution, determining the average step size. [...] We test our Bayesian learning algorithms on simulated data generated from a 2-dimensional isotropic kernel (σd = σ, ρd = ρ for d = 1, 2) using Gibbs sampling (Affandi et al., 2013a). We then learn the parameters under weakly informative inverse gamma priors on σ, ρ and α. Details are in the Supplement. [...] We place weakly informative inverse gamma priors on (α, ρ, σ), as specified in the Supplement, and learn the parameters using slice sampling with eigenvalue bounds as outlined in Sec. 3.3. |