Regret Bounds for Multilabel Classification in Sparse Label Regimes

Authors: Róbert Busa-Fekete, Heejin Choi, Krzysztof Dembczynski, Claudio Gentile, Henry Reeve, Balazs Szorenyi

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We fill the gap in the landscape of theoretical results by providing upper and lower finite-sample regret bounds in MLC with a focus on computational efficiency. We consider two learning setups, a nonparametric and a parametric one. ... As a last contribution, we derive MLC regret lower bounds for our MLC setups revealing that, at least in the non-parametric case, our upper bound for Hamming loss is optimal up to a log s factor, and that our regret upper bound for Precision@κ is optimal up to a log m factor. ... No experimential results.
Researcher Affiliation Collaboration Róbert Busa-Fekete Google Research busarobi@google.com Heejin Choi Google heejinc@google.com Krzysztof Dembczy nski Yahoo Research Poznan University of Technology kdembczynski@cs.put.poznan.pl Claudio Gentile Google Research cgentile@google.com Henry W. Reeve University of Bristol henry.reeve@bristol.ac.uk Balázs Szörényi Yahoo Research szorenyibalazs@gmail.com
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks.
Open Source Code No Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [N/A] No experimential results.
Open Datasets No No experimential results. ... No data provided.
Dataset Splits No No experimential results.
Hardware Specification No No experimential results.
Software Dependencies No The paper is theoretical and does not report any experiments that would require specific software dependencies with version numbers.
Experiment Setup No No experimential results.