# Aspect-augmented Adversarial Networks for Domain Adaptation

arXiv: https://arxiv.org/abs/1701.00188 · HTML: https://mlsanity.com/paper/1701.00188/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 1 Jan 2017.

## Built on by (most cited first, 12 of 12)

- [Multi-Source Domain Adaptation with Mixture of Experts](https://mlsanity.com/paper/1809.02256/) (2018)
- [Adversarial Learning for Zero-Shot Stance Detection on Social Media](https://mlsanity.com/paper/2105.06603/) (2021)
- [Deep Co-Attention Network for Multi-View Subspace Learning](https://mlsanity.com/paper/2102.07751/) (2021)
- [Robust Hybrid Learning With Expert Augmentation](https://mlsanity.com/paper/2202.03881/) (2022)
- [Disentangling Representations of Text by Masking Transformers](https://mlsanity.com/paper/2104.07155/) (2021)
- [A Data Fusion Framework for Multi-Domain Morality Learning](https://mlsanity.com/paper/2304.02144/) (2023)
- [DAME: Domain Adaptation for Matching Entities](https://mlsanity.com/paper/2204.09244/) (2022)
- [Graph Contrastive Learning with Generative Adversarial Network](https://mlsanity.com/paper/2308.00535/) (2023)
- [Adapting to the Long Tail: A Meta-Analysis of Transfer Learning Research for Language Understanding Tasks](https://mlsanity.com/paper/2111.01340/) (2021)
- [Comprehensive Attribution: Inherently Explainable Vision Model with Feature Detector](https://mlsanity.com/paper/2407.19308/) (2024)
- [Estimating Causal Effects of Text Interventions Leveraging LLMs](https://mlsanity.com/paper/2410.21474/) (2024)
- [Label Alignment Regularization for Distribution Shift](https://mlsanity.com/paper/2211.14960/) (2022)

## Abstract (arXiv)

We introduce a neural method for transfer learning between two (source and target) classification tasks or aspects over the same domain. Rather than training on target labels, we use a few keywords pertaining to source and target aspects indicating sentence relevance instead of document class labels. Documents are encoded by learning to embed and softly select relevant sentences in an aspect-dependent manner. A shared classifier is trained on the source encoded documents and labels, and applied to target encoded documents. We ensure transfer through aspect-adversarial training so that encoded documents are, as sets, aspect-invariant. Experimental results demonstrate that our approach outperforms different baselines and model variants on two datasets, yielding an improvement of 27% on a pathology dataset and 5% on a review dataset.