Entangled Kernels
Authors: Riikka Huusari, Hachem Kadri
IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We provide an empirical evaluation of EKL performance which demonstrates its effectiveness on artificial data as well as real benchmarks (Section 5). In this section the performance of our algorithm is illustrated with artificial and real datasets. |
| Researcher Affiliation | Academia | Riikka Huusari and Hachem Kadri Aix-Marseille University, CNRS, LIS, Marseille, France {riikka.huusari, hachem.kadri}@lis-lab.fr |
| Pseudocode | Yes | Algorithm 1 Entangled Kernel Learning (EKL) |
| Open Source Code | No | 3The code will be made available at RH’s personal webpage |
| Open Datasets | Yes | Concrete slump test (UCI dataset repository) with 103 data samples and three output variables; Sarcos5 is a dataset characterizing robot arm movements with 7 tasks; Weather6 has daily weather data (p = 365) from 35 stations. (Footnote 5: www.gaussianprocess.org/gpml/data/ and Footnote 6: https://www.psych.mcgill.ca/misc/fda/) |
| Dataset Splits | Yes | In all the experiments we cross-validate over various regularization parameters λ, and for EKL also γs controlling the combination of alignments. averaged over 10 data partitions. |
| Hardware Specification | No | The paper discusses computational efficiency and mentions general terms like 'GPU' in context of previous work, but it does not provide specific details about the hardware (e.g., CPU or GPU models, memory) used to run the experiments. |
| Software Dependencies | No | The paper mentions using 'OKL' code from authors [Dinuzzo et al., 2011] (footnote 4: http://people.tuebingen.mpg.de/fdinuzzo/okl.html) and that 'The optimization is performed on sphere manifold' using 'manopt.org/' and 'pymanopt.github.io/' (footnote 2). However, no specific version numbers are provided for these software tools or dependencies. |
| Experiment Setup | Yes | In all the experiments we cross-validate over various regularization parameters λ, and for EKL also γs controlling the combination of alignments. |