Boosted Histogram Transform for Regression
Authors: Yuchao Cai, Hanyuan Hang, Hanfang Yang, Zhouchen Lin
ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In the experiments, compared with other state-of-the-art algorithms such as gradient boosted regression tree (GBRT), Breiman s forest, and kernel-based methods, our BHTR algorithm shows promising performance on both synthetic and real datasets. |
| Researcher Affiliation | Collaboration | 1 AI Lab, Samsung Research China Beijing (SRC-B), Beijing, China 2 School of Statistics, Renmin University of China, Beijing, China 3 Key Lab. of Machine Perception (Mo E), School of EECS, Peking University, Beijing, China. |
| Pseudocode | Yes | Algorithm 1 Boosted Histogram Transform for Regression |
| Open Source Code | No | The paper does not provide any statement or link indicating the availability of the source code for the proposed methodology. |
| Open Datasets | No | The paper mentions using "three benchmark synthetic datasets and eight real datasets" (Table 2 lists names like ABA, MPG, etc.) and states "Further information can be found in the supplementary material" for real datasets. However, it does not provide direct links, DOIs, or formal citations with author/year for public access in the main text. |
| Dataset Splits | Yes | We perform experiments with n = 500 training data and then predict 2000 test observations... we perform experiments with 350 training data and 150 validation data... For each function, 1000 observations are generated for training and another 1000 are for testing. The cross-validation procedure is adopted for hyper-parameter selection. |
| Hardware Specification | No | The paper mentions 'High-performance Computing Platform of Renmin University of China' but does not specify any hardware details such as GPU/CPU models, memory, or other computational resources. |
| Software Dependencies | No | The paper mentions using "sklearn in python" and implies "numpy" through "np.logspace" but does not specify any version numbers for these software components. |
| Experiment Setup | Yes | We select ρ = 0.01 and T = 500. ...the learning rate ρ over the set {0.01 + 0.03k, k = 0, . . . , 33}. ...impose a gird of size 4 on learning rate ρ {0.2, 0.1, 0.05, 0.01}, a grid of size 3 on smin { 4, 3, 2} and a grid of size 2 on smax smin {1, 2}. For each element from the Cartesian product of these grids, we run BHTR with iteration times T = 3000. The optimal iteration times t T and optimal parameters ρ, smin and smax are chosen by 5-fold crossvalidation. |