Abstract:Multi-domain learning (MDL) refers to simultaneously constructing a model or a set of models on datasets collected from different domains. Conventional approaches emphasize domain-shared information extraction and domain-private information preservation, following the shared-private framework (SP models), which offers significant advantages over single-domain learning. However, the limited availability of annotated data in each domain considerably hinders the effectiveness of conventional supervised MDL approaches in real-world applications. In this paper, we introduce a novel method called multi-domain contrastive learning (MDCL) to alleviate the impact of insufficient annotations by capturing both semantic and structural information from both labeled and unlabeled data.Specifically, MDCL comprises two modules: inter-domain semantic alignment and intra-domain contrast. The former aims to align annotated instances of the same semantic category from distinct domains within a shared hidden space, while the latter focuses on learning a cluster structure of unlabeled instances in a private hidden space for each domain. MDCL is readily compatible with many SP models, requiring no additional model parameters and allowing for end-to-end training. Experimental results across five textual and image multi-domain datasets demonstrate that MDCL brings noticeable improvement over various SP models.Furthermore, MDCL can further be employed in multi-domain active learning (MDAL) to achieve a superior initialization, eventually leading to better overall performance.

MDL Transduction

Transformer-Based Multi-Source Domain Adaptation Without Source Data.

Locality sensitive discriminant transductive learning

A Transductive Learning Algorithm Based on Support Vector Machine

Boosting Vision-Language Models with Transduction

Transductive discriminative dictionary learning approach for zero-shot classification

Efficient Convex Relaxation for Transductive Support Vector Machine.

Adaptive Regularization for Transductive Support Vector Machine.

Learning from Multi-Dimensional Partial Labels.

Learning by Transduction

A New Minimum Divergence Approach To Discriminative Training

A New Transductive Support Vector Machine Approach To Text Categorization

Learning with Progressive Transductive Support Vector Machine

Multi-Domain Learning From Insufficient Annotations

DDxT: Deep Generative Transformer Models for Differential Diagnosis

Transductive Minimax Probability Machine.

Transfer Discriminative Logmaps

Grammar induction by MDL-based distributional classification

Minimum Divergence Based Discriminative Training.

On the Relation between Internal Language Model and Sequence Discriminative Training for Neural Transducers

MMDL-Based Data Augmentation with Domain Knowledge for Time Series Classification