# Same Same, But Different: Conditional Multi-Task Learning for Demographic-Specific Toxicity Detection

arXiv: https://arxiv.org/abs/2302.07372 · HTML: https://mlsanity.com/paper/2302.07372/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 29 of its 58 references that are in the map.
- First posted on arXiv 14 Feb 2023.

## Builds on (most cited first, 29 of 29)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [Equality of Opportunity in Supervised Learning](https://mlsanity.com/paper/1610.02413/) (2016)
- [WILDS: A Benchmark of in-the-Wild Distribution Shifts](https://mlsanity.com/paper/2012.07421/) (2020)
- [Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semantics](https://mlsanity.com/paper/1705.07115/) (2017)
- [GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks](https://mlsanity.com/paper/1711.02257/) (2017)
- [An Overview of Multi-Task Learning in Deep Neural Networks](https://mlsanity.com/paper/1706.05098/) (2017)
- [Focal Loss for Dense Object Detection](https://mlsanity.com/paper/1708.02002/) (2017)
- [Mitigating Unwanted Biases with Adversarial Learning](https://mlsanity.com/paper/1801.07593/) (2018)
- [End-to-End Multi-Task Learning with Attention](https://mlsanity.com/paper/1803.10704/) (2018)
- [Cross-stitch Networks for Multi-task Learning](https://mlsanity.com/paper/1604.03539/) (2016)
- [Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification](https://mlsanity.com/paper/1903.04561/) (2019)
- [Multi-Task Learning with Deep Neural Networks: A Survey](https://mlsanity.com/paper/2009.09796/) (2020)
- [HateCheck: Functional Tests for Hate Speech Detection Models](https://mlsanity.com/paper/2012.15606/) (2020)
- [UberNet: Training a `Universal' Convolutional Neural Network for Low-, Mid-, and High-Level Vision using Diverse Datasets and Limited Memory](https://mlsanity.com/paper/1609.02132/) (2016)
- [Reducing Gender Bias in Abusive Language Detection](https://mlsanity.com/paper/1808.07231/) (2018)
- [NDDR-CNN: Layerwise Feature Fusing in Multi-Task CNNs by Neural Discriminative Dimensionality Reduction](https://mlsanity.com/paper/1801.08297/) (2018)
- [Jury Learning: Integrating Dissenting Voices into Machine Learning Models](https://mlsanity.com/paper/2202.02950/) (2022)
- [FairBatch: Batch Selection for Model Fairness](https://mlsanity.com/paper/2012.01696/) (2020)
- [Fully-adaptive Feature Sharing in Multi-Task Networks with Applications in Person Attribute Classification](https://mlsanity.com/paper/1611.05377/) (2016)
- [Deep Multi-task Representation Learning: A Tensor Factorisation Approach](https://mlsanity.com/paper/1605.06391/) (2016)
- [A Joint Many-Task Model: Growing a Neural Network for Multiple NLP Tasks](https://mlsanity.com/paper/1611.01587/) (2016)
- [Many Task Learning with Task Routing](https://mlsanity.com/paper/1903.12117/) (2019)
- [Optimising Equal Opportunity Fairness in Model Training](https://mlsanity.com/paper/2205.02393/) (2022)
- [AngryBERT: Joint Learning Target and Emotion for Hate Speech Detection](https://mlsanity.com/paper/2103.11800/) (2021)
- [It Takes Nine to Smell a Rat: Neural Multi-Task Learning for Check-Worthiness Prediction](https://mlsanity.com/paper/1908.07912/) (2019)
- [Multi-Task Learning with Sentiment, Emotion, and Target Detection to Recognize Hate Speech and Offensive Language](https://mlsanity.com/paper/2109.10255/) (2021)
- [Exploiting Transformer-based Multitask Learning for the Detection of Media Bias in News Articles](https://mlsanity.com/paper/2211.03491/) (2022)
- [WLV-RIT at GermEval 2021: Multitask Learning with Transformers to Detect Toxic, Engaging, and Fact-Claiming Comments](https://mlsanity.com/paper/2108.00057/) (2021)

## Built on by (most cited first, 10 of 10)

- [Sensitivity, Performance, Robustness: Deconstructing the Effect of Sociodemographic Prompting](https://mlsanity.com/paper/2309.07034/) (2023)
- [A Critical Survey on Fairness Benefits of Explainable AI](https://mlsanity.com/paper/2310.13007/) (2023)
- [Human-centered NLP Fact-checking: Co-Designing with Fact-checkers using Matchmaking for AI](https://mlsanity.com/paper/2308.07213/) (2023)
- [Voices in a Crowd: Searching for Clusters of Unique Perspectives](https://mlsanity.com/paper/2407.14259/) (2024)
- [Hate Speech Detection with Generalizable Target-aware Fairness](https://mlsanity.com/paper/2406.00046/) (2024)
- [Algorithmic Fairness: A Tolerance Perspective](https://mlsanity.com/paper/2405.09543/) (2024)
- [Mitigating Label Bias via Decoupled Confident Learning](https://mlsanity.com/paper/2307.08945/) (2023)
- [Mapping the Potential of Explainable AI for Fairness Along the AI Lifecycle](https://mlsanity.com/paper/2404.18736/) (2024)
- [Fairness-Aware Multi-Group Target Detection in Online Discussion](https://mlsanity.com/paper/2407.11933/) (2024)
- [Bias-Aware Mislabeling Detection via Decoupled Confident Learning](https://mlsanity.com/paper/2507.07216/) (2025)

## Abstract (arXiv)

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.