Efficient and Scalable Multi-Task Regression on Massive Number of Tasks
Authors: Xiao He, Francesco Alesiani, Ammar Shaker3763-3770
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | On both synthetic and real-world datasets, the proposed CCMTL outperforms seven state-of-the-art (So A) multi-task learning methods in terms of prediction accuracy as well as computational efficiency. |
| Researcher Affiliation | Industry | Xiao He NEC Labs Europe Heidelberg, Germany xiao.he@neclab.eu Francesco Alesiani NEC Labs Europe Heidelberg, Germany francesco.alesiani@neclab.eu Ammar Shaker NEC Labs Europe Heidelberg, Germany ammar.shaker@neclab.eu |
| Pseudocode | Yes | Algorithm 1: CCMTL Input : {Xt, Yt} for t = {1, 2, ..., T}, λ Output: W = {W1, ..., WT } 1 for t 1 to T do 2 Solve Wt by Linear Regression on {Xt, Yt} 4 Construct k-nearest neighbor graph G on W; 5 while not converge do 6 Update L using Eq. (4); 7 Update W by solving Eq. (7) using CMG (Ioannis, Miller, and Tolliver 2011); 9 return W; |
| Open Source Code | No | The information is insufficient. The paper states 'We implement CCMTL2' and provides a URL (ccmtlaaai.neclab.eu), but this URL leads to a project page and not an explicit statement of code release or a direct link to a source code repository for the methodology described in the paper. |
| Open Datasets | Yes | We acquired the dataset from UCI repository (Dheeru and Karra Taniskidou 2017). [...] We use the dataset from Malsar package (Zhou, Chen, and Ye 2011b). |
| Dataset Splits | Yes | All these hyperparameters are selected by internal 5-fold cross validation grid search on the training data. [...] Table 4 presents the prediction error, RMSE, on the Syn dataset with the ratio of training samples ranging from 20% to 40%. |
| Hardware Specification | No | The information is insufficient. The paper states 'All methods are implemented in Matlab and evaluated on a single thread,' but does not provide specific details about the CPU, GPU, memory, or other hardware used for the experiments. |
| Software Dependencies | No | The information is insufficient. The paper mentions 'All methods are implemented in Matlab' and refers to the 'Snap VX python package', but does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | Yes | The number of nearest neighbors k is set to 10 to get the initial graph. [...] CCMTL, STL, ITL, L21, Trace and Fuse MTL need one hyperparameter that is selected from [10 5, 105]. [...] All these hyperparameters are selected by internal 5-fold cross validation grid search on the training data. |