Random Features for Shift-Invariant Kernels with Moment Matching
Authors: Weiwei Shen, Zhihui Yang, Jun Wang
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our extensive empirical studies and comparisons with several highly competitive peer methods verify the superiority of the proposed algorithm in Gram matrix approximation and generalization errors in regression. For validation, we provide detailed theoretical proofs and empirical comparisons with six state-of-the-art sampling methods across four standard benchmarks. |
| Researcher Affiliation | Collaboration | School of Computer Science and Software Engineering East China Normal University, Shanghai, China GE Global Research Center, Niskayuna, NY, USA |
| Pseudocode | Yes | Algorithm 1 Random Features with Moment Matching |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. |
| Open Datasets | Yes | Data: Four benchmark datasets with relatively high dimensions are examined in our experiments: (a) YP90 with 10000 and 2000 90-dimensional data points for training and testing, respectively; (b) QM274 with 6000 and 1165 274-dimensional data points for training and testing, respectively; (c) MNIST300 with 8000 and 2000 300-dimensional data points for training and testing, respectively; and (d) LR500 with 8000 and 2000 500-dimensional data points for training and testing, respectively. |
| Dataset Splits | Yes | Data: Four benchmark datasets with relatively high dimensions are examined in our experiments: (a) YP90 with 10000 and 2000 90-dimensional data points for training and testing, respectively; (b) QM274 with 6000 and 1165 274-dimensional data points for training and testing, respectively; (c) MNIST300 with 8000 and 2000 300-dimensional data points for training and testing, respectively; and (d) LR500 with 8000 and 2000 500-dimensional data points for training and testing, respectively. |
| Hardware Specification | No | The paper discusses computational costs but does not provide specific hardware details (e.g., exact GPU/CPU models, processor types, memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiment. |
| Experiment Setup | Yes | We specify the band width σ as the average distance of all data points to their tenth nearest neighbors unless otherwise stated. We first implement principal component decomposition on 4096 sampled random features. Then, we build principal component regression models by all 4096 principal components and by the first 512 principal components, respectively. |