Kernel Logistic Regression Approximation of an Understandable ReLU Neural Network
Authors: Marie Guyomard, Susana Barbosa, Lionel Fillatre
ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We evaluate the relevance of both the SATURNN model and its approximation methods δLR(x, η) in (15), δKLR(x, α) in (23) and δEKLR(x, α) in (32) on two simulated datasets and on one real one. ... Tables 1 and 2 summarize the average accuracies of each model computed from a 5-fold cross-validation on both training and test samples. |
| Researcher Affiliation | Academia | 1Laboratory I3S, University Cˆote d Azur, Sophia-Antipolis, France 2Laboratory IPMC, University Cˆote d Azur, Sophia-Antipolis, France. |
| Pseudocode | No | The paper describes methods mathematically but does not include any explicit pseudocode or algorithm blocks. |
| Open Source Code | No | The paper references third-party libraries like Scikit-Learn but does not provide an explicit statement or link to the authors' own open-source code for their described methodology. |
| Open Datasets | Yes | We chose to derive experiments using the Framingham dataset (Mahmood et al., 2014). |
| Dataset Splits | Yes | Tables 1 and 2 summarize the average accuracies of each model computed from a 5-fold cross-validation on both training and test samples. |
| Hardware Specification | No | The paper mentions 'Computation Time' in its results tables but does not provide specific hardware details such as GPU or CPU models used for running the experiments. |
| Software Dependencies | No | The paper mentions 'Scikit-Learn Logistic Regression' but does not specify its version number or other software dependencies with version information. |
| Experiment Setup | Yes | The β(0)k s satisfy β(0)k N(0, 1) for k = {0, . . . , p}. The thresholds b(0)k are initialized according to a uniform distribution on the interval [ r, r] since xi 2 r: b(0)k U[ r, +r]. A ℓ2-regularization is applied... The optimal number of spline basis for MARS or the optimal width and number of DTs composing RFs and EBMs are computed with gridsearch algorithms. |