Efficient Algorithms for Sum-Of-Minimum Optimization

Authors: Lisang Ding, Ziang Chen, Xinshang Wang, Wotao Yin

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The efficiency of our algorithms is numerically examined on multiple tasks, including generalized principal component analysis, mixed linear regression, and small-scale neural network training.
Researcher Affiliation Collaboration 1Department of Mathematics, University of California, Los Angeles, Los Angeles, CA, USA 2Department of Mathematics, Massachusetts Institute of Technology, Cambridge, MA, USA 3Decision Intelligence Lab, Alibaba US, Bellevue, WA, USA.
Pseudocode Yes The pseudo-code of this algorithm is shown in Algorithm 1.
Open Source Code Yes Our code with documentation can be found at https://github.com/Lisang Ding/ Sum-of-Minimum_Optimization.
Open Datasets No The dataset {(ai, bi)}N i=1 for the ℓ2-regularized mixed linear regression is synthetically generated in the following way:...
Dataset Splits No No specific dataset split information (e.g., percentages, sample counts for training, validation, and test sets) or mention of a validation set was provided. The paper states: "In our experiments, the training dataset size is N = 1000 and the testing dataset size is 200."
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory, or cloud instance types) used for running experiments are provided in the paper.
Software Dependencies No Software components like "ADAM optimizer" are mentioned, but no specific version numbers are provided to ensure reproducibility.
Experiment Setup Yes We set the maximum iteration number as 50 for Algorithm 2 with (5) and terminate the algorithm once the objective function stops decreasing... We use the default ADAM learning rate γ = 1e 3. We set r = 10 in Lloyd s Algorithm 2 and fix the cluster number k = 5.