Neural Interaction Transparency (NIT): Disentangling Learned Interactions for Improved Interpretability
Authors: Michael Tsang, Hanpeng Liu, Sanjay Purushotham, Pavankumar Murali, Yan Liu
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We validate the efficacy of our method first on synthetic data and then on four real-world datasets. [...] Table 3: Predictive performance of NIT. |
| Researcher Affiliation | Collaboration | 1University of Southern California 2IBM T.J. Watson Research Center |
| Pseudocode | No | The paper describes the method's steps but does not include structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide an explicit statement or link for open-source code for the described methodology. |
| Open Datasets | Yes | The real-world datasets (Table 2) include two regression datasets previously studied in statistical interaction research: Cal Housing [24] and Bike Sharing [6], and two binary classification datasets MIMIC-III [15] and CIFAR-10 binary. CIFAR-10 binary is a binary classification dataset (derived from CIFAR-10 [17]) |
| Dataset Splits | Yes | N = 3e4 at equal train/validation/test splits and different regularizations, which were tuned on the validation set. [...] For all real-world datasets except CIFAR-10 binary, 5-fold cross-validation was used, where model training was done on 3 folds, validation on the 4th fold and testing on the 5th. For the CIFAR dataset, the standard test set was only used for testing, and an inner 5-fold cross-validation was done on an 80%-20% train-validation split. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU, GPU models, or memory specifications) used for running experiments. |
| Software Dependencies | No | The paper mentions 'Re LU nonlinearity' and 'ADAM optimizer [16]' but does not provide specific software dependencies with version numbers (e.g., programming languages, libraries, or frameworks). |
| Experiment Setup | Yes | Learning rate was fixed at 5e 2 while the disentangling regularization was applied. For the hyperparameters of baselines and those not specific to NIT, tuning was done on the validation set. For all experiments with neural nets, we use the ADAM optimizer [16] and early stopping on validation sets. |