Does Preprocessing Help Training Over-parameterized Neural Networks?

Authors: Zhao Song, Shuo Yang, Ruizhe Zhang

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical From the technical perspective, our result is a sophisticated combination of tools in different fields, greedy-type convergence analysis in optimization, sparsity observation in practical work, high-dimensional geometric search in data structure, concentration and anti-concentration in probability. Our results also provide theoretical insights for a large number of previously established fast training methods. In addition, our classical algorithm can be generalized to the Quantum computation model. Interestingly, we can get a similar sublinear cost per iteration but avoid preprocessing initial weights or input data points.
Researcher Affiliation Collaboration Zhao Song Adobe Research zsong@adobe.com Shuo Yang The University of Texas at Austin yangshuo_ut@utexas.edu Ruizhe Zhang The University of Texas at Austin ruizhe@utexas.edu
Pseudocode Yes Algorithm 1 Half Space Report Data Structure, Algorithm 2 Training Neural Network via building a data structure of weights of the neural network, Algorithm 3 Training Neural Network via building a data-structure of the input data points
Open Source Code No Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? A: [N/A]
Open Datasets No The paper is theoretical and does not involve experiments with specific datasets. It refers to 'n data points in d-dimensional space' but does not specify a publicly available dataset or provide access information for any data used.
Dataset Splits No The paper is theoretical and does not include an experimental section with data splits (training, validation, test).
Hardware Specification No The paper is theoretical and does not conduct experiments, therefore no hardware specifications are mentioned.
Software Dependencies No The paper is theoretical and does not conduct experiments, therefore no specific software dependencies with version numbers are listed.
Experiment Setup No The paper is theoretical and does not contain an experimental section, therefore no specific experimental setup details or hyperparameters are provided.