The Many Shapley Values for Model Explanation
Authors: Mukund Sundararajan, Amir Najmi
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | While the bulk of this paper is axiomatic and theoretical, we will replicate some of our observations on a diabetes prediction task...We train a linear model using Scikit s implementation of Lasso regression (Tibshirani, 1996); we used the standard settings of the fitting algorithm and 75%-25% train-test split. The variance explained by the model is 35%. Figure 1 shows the distribution of attributions across 20 explicands for CES( ˆD)... |
| Researcher Affiliation | Industry | Mukund Sundararajan 1 Amir Najmi 1, 1Google LLC. Correspondence to: Mukund Sundararajan <mukunds@google.com>. |
| Pseudocode | Yes | Algorithm 1 Computing CES( ˆD) |
| Open Source Code | No | The paper does not provide an explicit statement or link for the open-sourcing of the code for the described methodology. |
| Open Datasets | Yes | We train our diabetes prediction models on a data set from the Scikit learning library (Pedregosa et al., 2011); this data set was originally used in (Efron et al., 2004). |
| Dataset Splits | No | The paper mentions a "75%-25% train-test split" but does not explicitly describe a separate validation split or how validation was performed. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions using "Scikit learning library" for training and "Scikit s implementation of Lasso regression" but does not provide specific version numbers for any software dependencies. |
| Experiment Setup | No | The paper states "we used the standard settings of the fitting algorithm" but does not provide specific details about hyperparameters, model initialization, or other training configurations. |