Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Geometric Intuition and Algorithms for Ev--SVM
Authors: Álvaro Barbero, Akiko Takeda, Jorge López
JMLR 2015 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We present now experimental results supporting our proposed ERCH model and the corresponding Rap Minos algorithm, as well as details on implementation. ... Figure 8 shows the obtained accuracy levels with Rap Minos for the range ν [0.1, 0.9] and a selection of class distances. |
| Researcher Affiliation | Academia | Alvaro Barbero EMAIL Department of Computer Science and Knowledge Engineering Institute Autonomous University of Madrid Madrid, Spain. Akiko Takeda EMAIL Department of Mathematical Informatics The University of Tokyo Tokyo, Japan. Jorge L opez EMAIL Department of Computer Science and Knowledge Engineering Institute Autonomous University of Madrid Madrid, Spain |
| Pseudocode | Yes | Algorithm 1 Subgradient Projection (SP) method for minx X f(x). ... Algorithm 2 Rap Minos method for ERCH-NPP |
| Open Source Code | Yes | A publicly available implementation of Rap Minos is provided. ... Project web page: https://bitbucket.org/albarji/rapminos . Source code and packages available. |
| Open Datasets | Yes | We now test the benefits provided by the augmented model capacity on real world data sets, obtained from the benchmark repository at R atsch (2000) ... Datasets available at http://www.raetschlab. org/Members/raetsch/benchmark. |
| Dataset Splits | Yes | In particular, we took 4/5 of the data set as training data and the remaining 1/5 as testing data. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running experiments. |
| Software Dependencies | No | The Rap Minos algorithm was implemented in Matlab... To solve the ERCH NPP in the non convex range we resorted to the presented Rap Minos method, while for the convex range we applied the standard ν SVM solver provided in LIBSVM (Chang and Lin, 2001). ... In our Rap Minos implementation we make use of the FISTA proximal algorithm (Beck and Teboulle, 2009)... we solve by making use of Matlab s internal LP solver routine linprog. The paper mentions software names like Matlab, LIBSVM, FISTA, and linprog, but does not provide specific version numbers for any of them. |
| Experiment Setup | Yes | Figure 8 shows the obtained accuracy levels with Rap Minos for the range ν [0.1, 0.9] and a selection of class distances. ... The training sample size and test sample size were set to m = 2 103 and em = 104, respectively, while the number of features was chosen as n = 10. ... We fixed p = 2 and chose ν as the one giving the highest validation performance, and for those data sets where ν was in the separable range, we chose a ν value slightly below the one for which hulls start intersecting. |