Matches in SemOpenAlex for { <https://semopenalex.org/work/W4380590974> ?p ?o ?g. }
Showing items 1 to 61 of
61
with 100 items per page.
- W4380590974 abstract "<p>In classification problems, mislabeled data can have a dramatic effect on the capability of a trained model. The traditional method of dealing with mislabeled data is through expert review of the data. However, this is not always ideal, due to both the large volume of data in many classification datasets, as with image datasets supporting deep learning models, and the limited availability of human experts for review of the data. Herein we propose an Ordered Sample Consensus (ORSAC) method to support data cleaning by flagging mislabeled data. This method is inspired by the Random Sample Consensus (RANSAC) method for outlier detection. In short, the method involves iteratively training and testing a model on different splits of the dataset, recording misclassifications, and flagging data which is frequently misclassified as probable mislabels. We evaluate the method by purposefully mislabeling subsets of the data and assessing the method’s capabilities to find such data. We demonstrate with three datasets, a mosquito image dataset, CIFAR-10, and CIFAR-100, that this method is reliable in finding mislabeled data with a high degree of accuracy. Our experimental results indicate a high proficiency of our methodology in identifying mislabeled data across these diverse datasets, with performance assessed using different mislabeling frequencies. </p>" @default.
- W4380590974 created "2023-06-15" @default.
- W4380590974 creator A5051366682 @default.
- W4380590974 creator A5056397221 @default.
- W4380590974 creator A5076050363 @default.
- W4380590974 date "2023-06-14" @default.
- W4380590974 modified "2023-09-30" @default.
- W4380590974 title "An Ordered Sample Consensus (ORSAC) Method for Data Cleaning Inspired by RANSAC: Identifying Probable Mislabeled Data" @default.
- W4380590974 doi "https://doi.org/10.36227/techrxiv.23511453" @default.
- W4380590974 hasPublicationYear "2023" @default.
- W4380590974 type Work @default.
- W4380590974 citedByCount "0" @default.
- W4380590974 crossrefType "posted-content" @default.
- W4380590974 hasAuthorship W4380590974A5051366682 @default.
- W4380590974 hasAuthorship W4380590974A5056397221 @default.
- W4380590974 hasAuthorship W4380590974A5076050363 @default.
- W4380590974 hasBestOaLocation W43805909741 @default.
- W4380590974 hasConcept C114744707 @default.
- W4380590974 hasConcept C115961682 @default.
- W4380590974 hasConcept C119857082 @default.
- W4380590974 hasConcept C124101348 @default.
- W4380590974 hasConcept C153180895 @default.
- W4380590974 hasConcept C154945302 @default.
- W4380590974 hasConcept C166957645 @default.
- W4380590974 hasConcept C185592680 @default.
- W4380590974 hasConcept C198531522 @default.
- W4380590974 hasConcept C205649164 @default.
- W4380590974 hasConcept C2777548347 @default.
- W4380590974 hasConcept C41008148 @default.
- W4380590974 hasConcept C43617362 @default.
- W4380590974 hasConcept C79337645 @default.
- W4380590974 hasConceptScore W4380590974C114744707 @default.
- W4380590974 hasConceptScore W4380590974C115961682 @default.
- W4380590974 hasConceptScore W4380590974C119857082 @default.
- W4380590974 hasConceptScore W4380590974C124101348 @default.
- W4380590974 hasConceptScore W4380590974C153180895 @default.
- W4380590974 hasConceptScore W4380590974C154945302 @default.
- W4380590974 hasConceptScore W4380590974C166957645 @default.
- W4380590974 hasConceptScore W4380590974C185592680 @default.
- W4380590974 hasConceptScore W4380590974C198531522 @default.
- W4380590974 hasConceptScore W4380590974C205649164 @default.
- W4380590974 hasConceptScore W4380590974C2777548347 @default.
- W4380590974 hasConceptScore W4380590974C41008148 @default.
- W4380590974 hasConceptScore W4380590974C43617362 @default.
- W4380590974 hasConceptScore W4380590974C79337645 @default.
- W4380590974 hasLocation W43805909741 @default.
- W4380590974 hasOpenAccess W4380590974 @default.
- W4380590974 hasPrimaryLocation W43805909741 @default.
- W4380590974 hasRelatedWork W1508595018 @default.
- W4380590974 hasRelatedWork W1993930080 @default.
- W4380590974 hasRelatedWork W2093014859 @default.
- W4380590974 hasRelatedWork W2185588454 @default.
- W4380590974 hasRelatedWork W2765767028 @default.
- W4380590974 hasRelatedWork W2898663890 @default.
- W4380590974 hasRelatedWork W2972372678 @default.
- W4380590974 hasRelatedWork W4306154256 @default.
- W4380590974 hasRelatedWork W4380562958 @default.
- W4380590974 hasRelatedWork W20692035 @default.
- W4380590974 isParatext "false" @default.
- W4380590974 isRetracted "false" @default.
- W4380590974 workType "article" @default.