A New PAC-Bayesian Perspective on Domain Adaptation

Authors: Pascal Germain, Amaury Habrard, François Laviolette, Emilie Morvant

ICML 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Then, we infer a learning algorithm and perform experiments on real data.
Researcher Affiliation Academia INRIA, SIERRA Project-Team, 75589, Paris, France, and D.I., Ecole Normale Superieure, 75230 Paris, France
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks. It describes the DALC algorithm in prose and mathematical equations.
Open Source Code No The paper does not include an unambiguous statement that the authors are releasing the code for the work described, nor does it provide a direct link to a source-code repository.
Open Datasets Yes Secondly, we evaluate DALC on the classical Amazon.com Reviews benchmark (Blitzer et al., 2006)
Dataset Splits Yes Each parameter is selected with a grid search thanks to a usual cross-validation (CV) on the source sample for SVM, and thanks to a reverse validation procedure11 (RCV) for CODA, DASVM, PBDA, and DALC.
Hardware Specification No The paper does not provide any specific hardware details (like GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper mentions 'scikit-learn' but does not provide specific version numbers for it or any other ancillary software dependencies.
Experiment Setup Yes For a source S={(xi, yi)}ms i=1 and a target T={(x i)}mt i=1 samples of potentially different size, and some hyperparameters C>0, B>0, minimizing the next objective function w.r.t w R is equivalent to minimize the above bound.