Learning Confidence Sets using Support Vector Machines
Authors: Wenbo Wang, Xingye Qiao
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 6 Numerical Studies, 6.1 Simulation, 6.2 Real Data Analysis, In the study, we use solver Cplex to solve the quadratic programming problem arising in CSVM. For other methods, we use existing R packages glmnet, gelnet, class, random Forest and e1071. We run the simulation multiple times (1,000 times for Example 1 and 100 times for Example 2 and 3) and report the average and standard error. |
| Researcher Affiliation | Academia | Wenbo Wang Department of Mathematical Sciences Binghamton University Binghamton, NY 13902 wang2@math.binghamton.edu Xingye Qiao* Department of Mathematical Sciences Binghamton University Binghamton, NY 13902 qiao@math.binghamton.edu |
| Pseudocode | No | The paper discusses algorithms and provides mathematical optimization formulations (e.g., equations (7) and (8)) but does not include structured pseudocode blocks or algorithms labeled as such. |
| Open Source Code | No | The paper does not contain any explicit statements about releasing source code for the described methodology or provide a link to a code repository. It only mentions using third-party solvers and R packages. |
| Open Datasets | Yes | We conduct the comparison on the hand-written zip code data [13]. |
| Dataset Splits | Yes | In each case, an independent tuning set with the same sample size as the training set is generated for parameter tuning. The testing set has 20000 observations (10000 or nearly 10000 for each class). The training and tuning data both have sample size 800, with 600 from class 1 and 200 from class 1 to preserve the unbalance nature of the data set. |
| Hardware Specification | No | The paper mentions software used (solver Cplex, R packages) but does not provide any specific hardware details such as CPU/GPU models or memory specifications used for the experiments. |
| Software Dependencies | No | The paper mentions using 'solver Cplex' and 'existing R packages glmnet, gelnet, class, randomForest and e1071' but does not specify their version numbers. |
| Experiment Setup | Yes | We search for the optimal ρ in the Gaussian kernel exp ( x y 2/ρ2) from the grid 10ˆ{ 0.5, 0.25, 0, 0.25, 0.5, 0.75, 1} and the optimal degree for polynomial kernel from {2, 3, 4}. For each fixed candidate hyper-parameter, we choose λ from a grid of candidate values ranging from 10 4 to 102 by the following two-step searching scheme. During training, we oversample class 1 by counting each observation three times to alleviate the unbalanced classes issue. |