Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386361503> ?p ?o ?g. }
Showing items 1 to 62 of
62
with 100 items per page.
- W4386361503 abstract "The internal workings of deep learning models are opaque and are considered as black boxes due to their nested and non-linear structure. This opaque nature of the deep neural networks makes it difficult to interpret the reason behind their output, thus reducing trust and verifiability of the system where these models are applied. This paper explains a systematic approach to identify the clusters with most misclassifications or false label annotations. For this research, we extracted the activation vectors from a deep learning model, DNABERT, and visualized them using t-SNE to decode the reason behind the results that are produced. We applied K-means in a hierarchical fashion on the activation vectors for a set of training instances. We analyzed cluster mean activation vectors to find any patterns in the errors across K-means clusters. The cluster analysis revealed that the predictions were uniform, or nearly 100 percent same, in clusters of similar activation vectors. It was found that two clusters containing most of their objects belonging to the same true class tend to be closer together than clusters of opposite classes. The means of objects of the same true label are closer if two clusters have the same predicted labels rather than opposite predicted labels, showing that the activation vectors reflect both predicted and true classes. We propose a heuristic to find the clusters with a high number of misclassifications or incorrect label annotations using the between clusters and within clusters mean vector analysis. This can aid in identifying misclassifications of DNA sequences or problems with sequence tagging." @default.
- W4386361503 created "2023-09-02" @default.
- W4386361503 creator A5009120802 @default.
- W4386361503 creator A5086066612 @default.
- W4386361503 date "2023-07-01" @default.
- W4386361503 modified "2023-09-27" @default.
- W4386361503 title "Finding BERT Errors Using Activation Vectors" @default.
- W4386361503 cites W2282821441 @default.
- W4386361503 cites W2525711135 @default.
- W4386361503 cites W2752194699 @default.
- W4386361503 cites W2963123635 @default.
- W4386361503 cites W2978934711 @default.
- W4386361503 cites W2985060393 @default.
- W4386361503 cites W3000716014 @default.
- W4386361503 cites W3082665562 @default.
- W4386361503 cites W3118485687 @default.
- W4386361503 cites W3127238141 @default.
- W4386361503 cites W3187290849 @default.
- W4386361503 cites W3214404698 @default.
- W4386361503 cites W4200296112 @default.
- W4386361503 cites W4226108798 @default.
- W4386361503 doi "https://doi.org/10.1109/bigdataservice58306.2023.00010" @default.
- W4386361503 hasPublicationYear "2023" @default.
- W4386361503 type Work @default.
- W4386361503 citedByCount "0" @default.
- W4386361503 crossrefType "proceedings-article" @default.
- W4386361503 hasAuthorship W4386361503A5009120802 @default.
- W4386361503 hasAuthorship W4386361503A5086066612 @default.
- W4386361503 hasConcept C153180895 @default.
- W4386361503 hasConcept C154945302 @default.
- W4386361503 hasConcept C164866538 @default.
- W4386361503 hasConcept C173801870 @default.
- W4386361503 hasConcept C177264268 @default.
- W4386361503 hasConcept C199360897 @default.
- W4386361503 hasConcept C2777212361 @default.
- W4386361503 hasConcept C41008148 @default.
- W4386361503 hasConcept C50644808 @default.
- W4386361503 hasConceptScore W4386361503C153180895 @default.
- W4386361503 hasConceptScore W4386361503C154945302 @default.
- W4386361503 hasConceptScore W4386361503C164866538 @default.
- W4386361503 hasConceptScore W4386361503C173801870 @default.
- W4386361503 hasConceptScore W4386361503C177264268 @default.
- W4386361503 hasConceptScore W4386361503C199360897 @default.
- W4386361503 hasConceptScore W4386361503C2777212361 @default.
- W4386361503 hasConceptScore W4386361503C41008148 @default.
- W4386361503 hasConceptScore W4386361503C50644808 @default.
- W4386361503 hasLocation W43863615031 @default.
- W4386361503 hasOpenAccess W4386361503 @default.
- W4386361503 hasPrimaryLocation W43863615031 @default.
- W4386361503 hasRelatedWork W2033914206 @default.
- W4386361503 hasRelatedWork W2042327336 @default.
- W4386361503 hasRelatedWork W2046077695 @default.
- W4386361503 hasRelatedWork W2146076056 @default.
- W4386361503 hasRelatedWork W2163831990 @default.
- W4386361503 hasRelatedWork W2378160586 @default.
- W4386361503 hasRelatedWork W2386387936 @default.
- W4386361503 hasRelatedWork W2996038082 @default.
- W4386361503 hasRelatedWork W3003836766 @default.
- W4386361503 hasRelatedWork W3047965787 @default.
- W4386361503 isParatext "false" @default.
- W4386361503 isRetracted "false" @default.
- W4386361503 workType "article" @default.