Peirou Liang

Yin Xie, Kaicheng Yang, Peirou Liang et al.

Large Multimodal Models (LMMs) often face a modality representation gap during pretraining: while language embeddings remain stable, visual representations are highly sensitive to contextual noise (e.g., background clutter). To address this issue, we introduce a visual comprehension stage, which we call ViCToR (Visual Comprehension via Token Reconstruction), a novel pretraining framework for LMMs. ViCToR employs a learnable visual token pool and utilizes the Hungarian matching algorithm to select semantically relevant tokens from this pool for visual token replacement. Furthermore, by integrating a visual token reconstruction loss with dense semantic supervision, ViCToR can learn tokens which retain high visual detail, thereby enhancing the large language model's (LLM's) understanding of visual information. After pretraining on 3 million publicly accessible images and captions, ViCToR achieves state-of-the-art results, improving over LLaVA-NeXT-8B by 10.4%, 3.2%, and 7.2% on the MMStar, SEED$^I$, and RealWorldQA benchmarks, respectively. Code is available at https://github.com/deepglint/Victor.

Haodi Zhang, Chenyu Xu, Peirou Liang et al.

Recent studies show that deep learning models achieve good performance on medical imaging tasks such as diagnosis prediction. Among the models, multimodality has been an emerging trend, integrating different forms of data such as chest X-ray (CXR) images and electronic medical records (EMRs). However, most existing methods incorporate them in a model-free manner, which lacks theoretical support and ignores the intrinsic relations between different data sources. To address this problem, we propose a knowledge-driven and data-driven framework for lung disease diagnosis. By incorporating domain knowledge, machine learning models can reduce the dependence on labeled data and improve interpretability. We formulate diagnosis rules according to authoritative clinical medicine guidelines and learn the weights of rules from text data. Finally, a multimodal fusion consisting of text and image data is designed to infer the marginal probability of lung disease. We conduct experiments on a real-world dataset collected from a hospital. The results show that the proposed method outperforms the state-of-the-art multimodal baselines in terms of accuracy and interpretability.