PCA-based Multi-Task Learning: a Random Matrix Approach

Authors: Malik Tiomoko, Romain Couillet, Frederic Pascal

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Supporting experiments on synthetic and real data benchmarks show that the proposed method achieves comparable performance with state-of-the-art MTL methods but at a significantly reduced computational cost.
Researcher Affiliation Collaboration 1Huawei Noah s Ark Lab, Paris, France 2LIG-Lab, Université de Grenoble Alpes, France 3L2S Centrale-Supélec, France.
Pseudocode Yes Algorithm 1. Proposed multi-class MTL SPCA algorithm.
Open Source Code Yes The proofs and Matlab codes to reproduce our main results and simulations, along with theoretical extensions and additional supporting results, are provided in the supplementary material.
Open Datasets Yes We here compare the performance of Algorithm 1 (MTL SPCA), on both synthetic and real data benchmarks... Image Clef dataset (Ionescu et al., 2017) ... Amazon review (textual) dataset8 (Blitzer et al., 2007) and the MNIST (image) dataset (Deng, 2012).
Dataset Splits Yes Figure 2. (Left) Theoretical (Th)/empirical (Emp) error rate for 2-class Gaussian mixture transfer with means µ1 = e[p] 1 , µ 1 = e[p] p , µ2 = βµ1 + p 1 β2µ 1 , p = 100, n1j = 1 000, n2j = 50; (Right) running time comparison (in sec); n = 2p, ntj/n = 0.25.
Hardware Specification No The paper mentions experiments were run 'on a modern laptop' but does not provide specific hardware details such as CPU or GPU models, or memory specifications.
Software Dependencies No The paper mentions 'Matlab codes' and 'Mex files' but does not provide specific version numbers for Matlab or any other software dependencies.
Experiment Setup Yes Figure 2. (Left) Theoretical (Th)/empirical (Emp) error rate for 2-class Gaussian mixture transfer with means µ1 = e[p] 1 , µ 1 = e[p] p , µ2 = βµ1 + p 1 β2µ 1 , p = 100, n1j = 1 000, n2j = 50; (Right) running time comparison (in sec); n = 2p, ntj/n = 0.25.