SnapBoost: A Heterogeneous Boosting Machine
Authors: Thomas Parnell, Andreea Anghel, Małgorzata Łazuka, Nikolas Ioannou, Sebastian Kurella, Peshal Agarwal, Nikolaos Papandreou, Haralampos Pozidis
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we present experimental results, using Open ML and Kaggle datasets, that show that Snap Boost is able to achieve better generalization loss than competing boosting frameworks, without taking significantly longer to tune. |
| Researcher Affiliation | Collaboration | Thomas Parnell IBM Research Zürich, Switzerland tpa@zurich.ibm.com Andreea Anghel IBM Research Zürich, Switzerland aan@zurich.ibm.com Małgorzata Łazuka ETH Zürich Zürich, Switzerland lazukam@student.ethz.ch Nikolas Ioannou IBM Research Zürich, Switzerland nio@zurich.ibm.com Sebastian Kurella ETH Zürich Zürich, Switzerland kurellas@student.ethz.ch Peshal Agarwal ETH Zürich Zürich, Switzerland agarwalp@student.ethz.ch Nikolaos Papandreou IBM Research Zürich, Switzerland npo@zurich.ibm.com Haralampos Pozidis IBM Research Zürich, Switzerland hap@zurich.ibm.com |
| Pseudocode | Yes | Algorithm 1 Heterogeneous Newton Boosting Machine |
| Open Source Code | No | The paper describes the implementation of Snap Boost but does not provide a link to its source code or explicitly state that it is open source. |
| Open Datasets | Yes | We compare XGBoost, Light GBM and Snap Boost across 10 binary classification datasets sourced from the Open ML platform [45]. [...] we use 3 datasets from the Kaggle platform. |
| Dataset Splits | Yes | 3x3 nested stratified cross-validation was used to perform hyper-parameter tuning and to obtain a reliable estimate of the generalization loss. [...] Since these datasets are relatively large, we perform a single train/validation/test split. |
| Hardware Specification | Yes | The results in this section were obtained using a multi-socket server with two 20-core Intel(R) Xeon(R) Gold 6230 CPUs @2.10GHz, 256 Gi B RAM, running Ubuntu 18.04. |
| Software Dependencies | Yes | We used XGBoost v1.1.0, Light GBM v2.3.1, Cat Boost v.0.23.2 and KTBoost v0.1.13. |
| Experiment Setup | Yes | Details regarding the SH implementation and the hyper-parameter ranges can be found in Appendix C and Appendix D respectively. |