TY - GEN
T1 - Same Same, But Different
T2 - 32nd ACM World Wide Web Conference, WWW 2023
AU - Gupta, Soumyajit
AU - Lee, Sooyong
AU - De-Arteaga, Maria
AU - Lease, Matthew
N1 - Publisher Copyright:
© 2023 Owner/Author.
PY - 2023/4/30
Y1 - 2023/4/30
N2 - Algorithmic bias often arises as a result of differential subgroup validity, in which predictive relationships vary across groups. For example, in toxic language detection, comments targeting different demographic groups can vary markedly across groups. In such settings, trained models can be dominated by the relationships that best fit the majority group, leading to disparate performance. We propose framing toxicity detection as multi-task learning (MTL), allowing a model to specialize on the relationships that are relevant to each demographic group while also leveraging shared properties across groups. With toxicity detection, each task corresponds to identifying toxicity against a particular demographic group. However, traditional MTL requires labels for all tasks to be present for every data point. To address this, we propose Conditional MTL (CondMTL), wherein only training examples relevant to the given demographic group are considered by the loss function. This lets us learn group specific representations in each branch which are not cross contaminated by irrelevant labels. Results on synthetic and real data show that using CondMTL improves predictive recall over various baselines in general and for the minority demographic group in particular, while having similar overall accuracy.
AB - Algorithmic bias often arises as a result of differential subgroup validity, in which predictive relationships vary across groups. For example, in toxic language detection, comments targeting different demographic groups can vary markedly across groups. In such settings, trained models can be dominated by the relationships that best fit the majority group, leading to disparate performance. We propose framing toxicity detection as multi-task learning (MTL), allowing a model to specialize on the relationships that are relevant to each demographic group while also leveraging shared properties across groups. With toxicity detection, each task corresponds to identifying toxicity against a particular demographic group. However, traditional MTL requires labels for all tasks to be present for every data point. To address this, we propose Conditional MTL (CondMTL), wherein only training examples relevant to the given demographic group are considered by the loss function. This lets us learn group specific representations in each branch which are not cross contaminated by irrelevant labels. Results on synthetic and real data show that using CondMTL improves predictive recall over various baselines in general and for the minority demographic group in particular, while having similar overall accuracy.
KW - conditional loss
KW - differential subgroup validity
KW - Multi-task learning
UR - https://www.scopus.com/pages/publications/85159265924
U2 - 10.1145/3543507.3583290
DO - 10.1145/3543507.3583290
M3 - Conference contribution
AN - SCOPUS:85159265924
T3 - ACM Web Conference 2023 - Proceedings of the World Wide Web Conference, WWW 2023
SP - 3689
EP - 3700
BT - ACM Web Conference 2023 - Proceedings of the World Wide Web Conference, WWW 2023
PB - Association for Computing Machinery, Inc
CY - New York
Y2 - 30 April 2023 through 4 May 2023
ER -