The Landscape of Non-convex Empirical Risk with Degenerate Population Risk

Authors: Shuang Li, Gongguo Tang, Michael B. Wakin

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We first conduct numerical experiments on the two examples introduced in Section 1, i.e., the rank-1 matrix sensing and phase retrieval problems. In both problems, we fix N = 2 and set x = [1 1] . Then, we generate the population risk and empirical risk based on the formulation introduced in these two examples. The contour plots of the population risk and a realization of empirical risk with M = 3 and M = 10 are given in Figure 4 for rank-1 matrix sensing and Figure 5 for phase retrieval. Next, we conduct another experiment on general-rank matrix sensing with k = 2, r = 3, N = 8, and a variety of M. We set U as the first r columns of an N N identity matrix and create X = U U . The population and empirical risks are then generated according to the model introduced in Section 3.1. As shown in Figure 6, the distance (averaged over 100 trials) between the local minima of the population and empirical risk decreases as we increase M.
Researcher Affiliation Academia Shuang Li, Gongguo Tang, and Michael B. Wakin Department of Electrical Engineering Colorado School of Mines Golden, CO 80401 {shuangli,gtang,mwakin}@mines.edu
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any statements regarding the release of open-source code or links to a code repository.
Open Datasets No The paper describes generating measurements and random matrices/vectors based on Gaussian distributions (e.g., 'Gaussian sensing matrix', 'Gaussian random vector') for numerical simulations, but does not specify or provide access information for any publicly available or open datasets.
Dataset Splits No The paper describes conducting numerical simulations with M samples but does not provide specific details on how data was split into training, validation, or test sets.
Hardware Specification No The paper does not provide any specific hardware details (e.g., CPU, GPU models, or memory specifications) used for running its experiments.
Software Dependencies No The paper does not provide any specific software names with version numbers for reproducibility.
Experiment Setup Yes In both problems, we fix N = 2 and set x = [1 1] . Then, we generate the population risk and empirical risk based on the formulation introduced in these two examples. The contour plots of the population risk and a realization of empirical risk with M = 3 and M = 10 are given in Figure 4 for rank-1 matrix sensing and Figure 5 for phase retrieval. Next, we conduct another experiment on general-rank matrix sensing with k = 2, r = 3, N = 8, and a variety of M.