Distributed Learning of Conditional Quantiles in the Reproducing Kernel Hilbert Space

Authors: Heng Lian

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Section 3 contains a simple numerical illustration, which is not used to verify the learning rate (which is difficult if not impossible) but just to illustrate the divide-and-conquer method can work reasonably. ... The simulation results are shown in Table 1 for τ = 0.5. We can examine the table in several ways. ... The estimation errors for different pairs of (n, m) when τ = 0.5.
Researcher Affiliation Academia Heng Lian City University of Hong Kong Shenzhen Research Institute, Shenzhen, China and Department of Mathematics, City University of Hong Kong, Hong Kong, China henglian@cityu.edu.hk
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any statements about providing open-source code for the described methodology, nor does it provide a link to a code repository.
Open Datasets No The paper uses synthetic data generated from the model yi = f0(xi) + (1 + xi)σ(ϵi Φ 1(τ)), where xi are generated uniformly on [0, 1], ϵi N(0, 1), with f0(x) = sin(2πx) and σ = 0.5. No publicly available dataset is used or provided.
Dataset Splits No The paper mentions that the tuning parameter λ is chosen 'to minimize the errors on independently generated test data'. However, it does not specify explicit training/validation/test splits (e.g., percentages or counts) for the main experiment or use cross-validation for model evaluation.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used to run the experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes The tuning parameter λ is chosen to minimize the errors on independently generated test data. For the RKHS, we use the Sobolev space of the second order. The sample is generated from the model yi = f0(xi) + (1 + xi)σ(ϵi Φ 1(τ)), where we set f0(x) = sin(2πx) and σ = 0.5. The simulations are carried out for different combinations of n {32, 64, 128, 256, 512, 1024} and m {1, 2, 4, 8, 16, 32}.