Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1]
Learning outside the Black-Box: The pursuit of interpretable models
Authors: Jonathan Crabbe, Yao Zhang, William Zame, Mihaela van der Schaar
NeurIPS 2020 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Using five familiar data sets from the UCI repository and two familiar machine learning algorithms, we demonstrate that our algorithm produces global interpretations that are both highly accurate and parsimonious (involve a small number of terms). |
| Researcher Affiliation | Academia | Jonathan Crabbe University of Cambridge EMAIL Yao Zhang University of Cambridge EMAIL William R. Zame University of California Los Angeles EMAIL Mihaela van der Schaar University of Cambridge University of California Los Angeles The Alan Turing Institute EMAIL |
| Pseudocode | Yes | The pseudo code of Symbolic Pursuit is provided in Section 5 of the supplementary material |
| Open Source Code | Yes | The code for Symbolic Pursuit is available at https://bitbucket.org/mvdschaar/mlforhealthlabpub and https://github.com/Jonathan Crabbe/Symbolic-Pursuit. |
| Open Datasets | Yes | We evaluate the performance of Symbolic Pursuit on two popular black-box models a Multilayer Perceptron (MLP) and Support Vector Machine (SVM) applied to five UCI datasets [12] including Wine Quality Red (Wine) and Yacht Hydrodynamics (Yacht), Boston Housing (Boston), Energy Efficiency (Energy) and Concrete Strength (Concrete). |
| Dataset Splits | No | For each experiment, we split the dataset into a training set (80%) and a test set (20%). No explicit mention of a separate validation split was found. |
| Hardware Specification | No | No specific hardware details (e.g., CPU, GPU models, or memory) used for running the experiments were provided. |
| Software Dependencies | No | Both models are implemented using the scikit-learn library [6] with the default hyperparameters. |
| Experiment Setup | No | Both models are implemented using the scikit-learn library [6] with the default hyperparameters. (This refers to the black-box models, not the Symbolic Pursuit algorithm's hyperparameters. No specific hyperparameters or training details are given for Symbolic Pursuit itself in the main text). |