Learning Sparse Task Relations in Multi-Task Learning
Authors: Yu Zhang, Qiang Yang
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on synthetic and real-world datasets demonstrate the effectiveness of the proposed method.In this section, we empirically test the performance of the proposed SPATS method. |
| Researcher Affiliation | Academia | Yu Zhang, Qiang Yang Department of Computer Science and Engineering Hong Kong University of Science and Technology {zhangyu,qyang}@cse.ust.hk |
| Pseudocode | No | The paper describes an 'Optimization Procedure' and refers to the FISTA algorithm, but it does not provide a structured pseudocode block or a clearly labeled algorithm box. |
| Open Source Code | No | The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | Yes | Five benchmark datasets, including School, Parkinson, Sentiment, Landmine and MHC-I datasets, are used in the experiment. The School dataset contains examination scores of 15362 students from 139 secondary schools in London during years 1985, 1986 and 1987, hence, there are totally 139 tasks. The input consists of the year of the examination, four school-specific and three student-specific attributes. Following (Evgeniou, Micchelli, and Pontil 2005), we replace each categorical attribute with one binary variable for each possible attribute value and as a result, there are 27 input attributes. |
| Dataset Splits | Yes | For each task, we generate 50 data points for training, 50 data points for validation to choose the regularization parameter with the set of candidate values as {0.001, 0.005, 0.01, 0.05, 0.1, 0.5, 1}, 100 data points for testing. For all the datasets, we randomly choose 20% data for training, 20% for validation, and the rest for testing. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running the experiments (e.g., CPU, GPU models, or memory specifications). |
| Software Dependencies | No | The paper refers to algorithms like FISTA but does not list any specific software dependencies or libraries with version numbers required to replicate the experiments. |
| Experiment Setup | Yes | The validation set is to choose the regularization parameters in all of the methods in comparison and the set of the candidate values for the regularization parameters is {0.001, 0.005, 0.01, 0.05, 0.1, 0.5, 1}.In order to solve problem (10), we use an alternating method which first optimizes problem (10) with respect to W and b by fixing Ω and then solves problem (10) with respect to Ω given the fixed W and b. |