Matches in SemOpenAlex for { <https://semopenalex.org/work/W4281395811> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W4281395811 abstract "Adversarial training has recently emerged as an important defense mechanism to robustify machine learning models in the presence adversarial examples. Although adversarial training can boost the robustness of machine learning algorithms by a margin, research has not been conducted to determine if adversarial training is effective in the long-term. As deployments of machine learning algorithms are characterized by dynamics, change of the underlying model is inevitable. The dynamics are a result of model's evolution over time by introducing new training data and drifting the model by changing its parameters. In this paper, we examine the limitations of adversarial training due to the temporal changes of machine learning models. Using a natural language task, we conduct various experiments using a variety of datasets to measure the impact of concept drift on the efficacy of adversarial training. In particular, our analysis shows that certain adversarially-trained models are even more prone to the drift than others. In particular, WordCNN and LSTM-based models are shown more susceptible to the temporal changes than others such as BERT. We validate our findings using multiple real-world datasets on different network architectures. Our work calls for further research into the temporal aspects of adversarial training." @default.
- W4281395811 created "2022-05-25" @default.
- W4281395811 creator A5000650499 @default.
- W4281395811 creator A5026101339 @default.
- W4281395811 creator A5077402873 @default.
- W4281395811 creator A5079230730 @default.
- W4281395811 date "2022-05-30" @default.
- W4281395811 modified "2023-09-30" @default.
- W4281395811 title "Quantifying the Performance of Adversarial Training on Language Models with Distribution Shifts" @default.
- W4281395811 cites W1981328081 @default.
- W4281395811 cites W2885195348 @default.
- W4281395811 cites W2963857521 @default.
- W4281395811 cites W2965690110 @default.
- W4281395811 cites W2983042735 @default.
- W4281395811 cites W2996851481 @default.
- W4281395811 cites W3120254979 @default.
- W4281395811 cites W3129221863 @default.
- W4281395811 cites W3164045934 @default.
- W4281395811 doi "https://doi.org/10.1145/3494108.3522764" @default.
- W4281395811 hasPublicationYear "2022" @default.
- W4281395811 type Work @default.
- W4281395811 citedByCount "0" @default.
- W4281395811 crossrefType "proceedings-article" @default.
- W4281395811 hasAuthorship W4281395811A5000650499 @default.
- W4281395811 hasAuthorship W4281395811A5026101339 @default.
- W4281395811 hasAuthorship W4281395811A5077402873 @default.
- W4281395811 hasAuthorship W4281395811A5079230730 @default.
- W4281395811 hasConcept C104317684 @default.
- W4281395811 hasConcept C119857082 @default.
- W4281395811 hasConcept C121332964 @default.
- W4281395811 hasConcept C127413603 @default.
- W4281395811 hasConcept C137293760 @default.
- W4281395811 hasConcept C153294291 @default.
- W4281395811 hasConcept C154945302 @default.
- W4281395811 hasConcept C185592680 @default.
- W4281395811 hasConcept C201995342 @default.
- W4281395811 hasConcept C2777211547 @default.
- W4281395811 hasConcept C2780451532 @default.
- W4281395811 hasConcept C37736160 @default.
- W4281395811 hasConcept C41008148 @default.
- W4281395811 hasConcept C55493867 @default.
- W4281395811 hasConcept C60777511 @default.
- W4281395811 hasConcept C63479239 @default.
- W4281395811 hasConcept C774472 @default.
- W4281395811 hasConcept C89198739 @default.
- W4281395811 hasConceptScore W4281395811C104317684 @default.
- W4281395811 hasConceptScore W4281395811C119857082 @default.
- W4281395811 hasConceptScore W4281395811C121332964 @default.
- W4281395811 hasConceptScore W4281395811C127413603 @default.
- W4281395811 hasConceptScore W4281395811C137293760 @default.
- W4281395811 hasConceptScore W4281395811C153294291 @default.
- W4281395811 hasConceptScore W4281395811C154945302 @default.
- W4281395811 hasConceptScore W4281395811C185592680 @default.
- W4281395811 hasConceptScore W4281395811C201995342 @default.
- W4281395811 hasConceptScore W4281395811C2777211547 @default.
- W4281395811 hasConceptScore W4281395811C2780451532 @default.
- W4281395811 hasConceptScore W4281395811C37736160 @default.
- W4281395811 hasConceptScore W4281395811C41008148 @default.
- W4281395811 hasConceptScore W4281395811C55493867 @default.
- W4281395811 hasConceptScore W4281395811C60777511 @default.
- W4281395811 hasConceptScore W4281395811C63479239 @default.
- W4281395811 hasConceptScore W4281395811C774472 @default.
- W4281395811 hasConceptScore W4281395811C89198739 @default.
- W4281395811 hasFunder F4320322120 @default.
- W4281395811 hasLocation W42813958111 @default.
- W4281395811 hasOpenAccess W4281395811 @default.
- W4281395811 hasPrimaryLocation W42813958111 @default.
- W4281395811 hasRelatedWork W2903483905 @default.
- W4281395811 hasRelatedWork W2952919291 @default.
- W4281395811 hasRelatedWork W3035729345 @default.
- W4281395811 hasRelatedWork W3193857078 @default.
- W4281395811 hasRelatedWork W3208304128 @default.
- W4281395811 hasRelatedWork W4286235935 @default.
- W4281395811 hasRelatedWork W4287549955 @default.
- W4281395811 hasRelatedWork W4304192736 @default.
- W4281395811 hasRelatedWork W4306177400 @default.
- W4281395811 hasRelatedWork W4311734044 @default.
- W4281395811 isParatext "false" @default.
- W4281395811 isRetracted "false" @default.
- W4281395811 workType "article" @default.