Semi-Supervised Sparse Gaussian Classification: Provable Benefits of Unlabeled Data

Authors: Eyar Azar, Boaz Nadler

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

Reproducibility Variable Result LLM Response
Research Type Experimental We present simulations that complement our theoretical analysis.
Researcher Affiliation Academia Eyar Azar Weizmann Institute of Science eyar.azar@weizmann.ac.il Boaz Nadler Weizmann Institute of Science boaz.nadler@weizmann.ac.il
Pseudocode Yes Algorithm 1 LSPCA
Open Source Code No The paper does not provide concrete access to source code (specific repository link, explicit code release statement, or code in supplementary materials) for the methodology described in this paper.
Open Datasets No The paper describes generating data according to a model: 'We generate L + n labeled and unlabeled samples according to the model (3)'. It does not refer to or provide access to a pre-existing public dataset.
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning. Data is generated per experiment.
Hardware Specification Yes All experiments were run on an Intel i7 CPU 2.10 GHz.
Software Dependencies No The paper mentions methods like 'iteratively proxy update (IPU)', 'ASPCA', and 'LSDF' but does not provide specific version numbers for any software dependencies, libraries, or frameworks used.
Experiment Setup Yes We run our SSL schemes with β = β (β (1 γα))/4 which satisfies the requirements of Theorem 3.2. We present experiments with p = 105, k = p0.4 = 100 and λ = 3. In the experiments we used Γ = 0.8, which gave the best performance.