Latent-Optimized Adversarial Neural Transfer for Sarcasm Detection
arXiv:2104.09261
Abstract
The existence of multiple datasets for sarcasm detection prompts us to apply transfer learning to exploit their commonality. The adversarial neural transfer (ANT) framework utilizes multiple loss terms that encourage the source-domain and the target-domain feature distributions to be similar while optimizing for domain-specific performance. However, these objectives may be in conflict, which can lead to optimization difficulties and sometimes diminished transfer. We propose a generalized latent optimization strategy that allows different losses to accommodate each other and improves training dynamics. The proposed method outperforms transfer learning and meta-learning baselines. In particular, we achieve 10.02% absolute performance gain over the previous state of the art on the iSarcasm dataset.
arXiv comment14 pages, 5 figures, published at NAACL-HLT 2021 conference, see https://www.aclweb.org/anthology/2021.naacl-main.425/
References in corpus (10)
- Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
- Natural Language Processing (almost) from Scratch
- Unrolled Generative Adversarial Networks
- Transfer Learning for Sequence Tagging with Hierarchical Recurrent Networks
- The Mechanics of n-Player Differentiable Games
- Multi-Task Learning for Sequence Tagging: An Empirical Study
- Deep Compressed Sensing
- LOGAN: Latent Optimisation for Generative Adversarial Networks
- Global Convergence to the Equilibrium of GANs using Variational Inequalities
- A Tight and Unified Analysis of Gradient-Based Methods for a Whole Spectrum of Games