Learning Using Generated Privileged Information by Text-to-Image Diffusion Models
This work addresses the problem of improving text classification accuracy for researchers and practitioners by using generated data as a cost-free enhancement during training, though it is incremental as it builds on existing privileged information and distillation methods.
The paper tackles the lack of available privileged information in knowledge distillation by generating synthetic images from text using diffusion models to train multimodal teachers, which then distill knowledge into a unimodal student for text classification, achieving noticeable performance gains on four datasets.
Learning Using Privileged Information is a particular type of knowledge distillation where the teacher model benefits from an additional data representation during training, called privileged information, improving the student model, which does not see the extra representation. However, privileged information is rarely available in practice. To this end, we propose a text classification framework that harnesses text-to-image diffusion models to generate artificial privileged information. The generated images and the original text samples are further used to train multimodal teacher models based on state-of-the-art transformer-based architectures. Finally, the knowledge from multimodal teachers is distilled into a text-based (unimodal) student. Hence, by employing a generative model to produce synthetic data as privileged information, we guide the training of the student model. Our framework, called Learning Using Generated Privileged Information (LUGPI), yields noticeable performance gains on four text classification data sets, demonstrating its potential in text classification without any additional cost during inference.