Advancing Multimodal Large Language Models in Chart Question Answering with Visualization-Referenced Instruction Tuning

被引:0
作者
Zeng, Xingchen [1 ]
Lin, Haichuan [1 ]
Ye, Yilin [1 ,2 ]
Zeng, Wei [1 ,2 ]
机构
[1] Hong Kong Univ Sci & Technol, Guangzhou, Peoples R China
[2] Hong Kong Univ Sci & Technol, Hong Kong, Peoples R China
基金
中国国家自然科学基金;
关键词
Chart-question answering; multimodal large language models; benchmark; OF-THE-ART;
D O I
暂无
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Emerging multimodal large language models (MLLMs) exhibit great potential for chart question answering (CQA). Recent efforts primarily focus on scaling up training datasets (i.e., charts, data tables, and question-answer (QA) pairs) through data collection and synthesis. However, our empirical study on existing MLLMs and CQA datasets reveals notable gaps. First, current data collection and synthesis focus on data volume and lack consideration of fine-grained visual encodings and QA tasks, resulting in unbalanced data distribution divergent from practical CQA scenarios. Second, existing work follows the training recipe of the base MLLMs initially designed for natural images, under-exploring the adaptation to unique chart characteristics, such as rich text elements. To fill the gap, we propose a visualization-referenced instruction tuning approach to guide the training dataset enhancement and model development. Specifically, we propose a novel data engine to effectively filter diverse and high-quality data from existing datasets and subsequently refine and augment the data using LLM-based generation techniques to better align with practical QA tasks and visual encodings. Then, to facilitate the adaptation to chart characteristics, we utilize the enriched data to train an MLLM by unfreezing the vision encoder and incorporating a mixture-of-resolution adaptation strategy for enhanced fine-grained recognition. Experimental results validate the effectiveness of our approach. Even with fewer training examples, our model consistently outperforms state-of-the-art CQA models on established benchmarks. We also contribute a dataset split as a benchmark for future research. Source codes and datasets of this paper are available at https://github.com/zengxingchen/ChartQA-MLLM.
引用
收藏
页码:525 / 535
页数:11
相关论文
共 76 条
[1]  
Abbas A, 2023, Arxiv, DOI [arXiv:2303.09540, 10.48550/arXiv.2303.09540]
[2]  
Achiam J., 2023, GPT 4 TECHNICAL REPO, DOI DOI 10.48550/ARXIV.2303.08774
[3]  
[Anonymous], 2024, OpenAI.gpt-4-vision-preview
[4]   VQA: Visual Question Answering [J].
Antol, Stanislaw ;
Agrawal, Aishwarya ;
Lu, Jiasen ;
Mitchell, Margaret ;
Batra, Dhruv ;
Zitnick, C. Lawrence ;
Parikh, Devi .
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, :2425-2433
[5]  
Bai JZ, 2023, Arxiv, DOI [arXiv:2308.12966, 10.48550/arXiv.2308.12966]
[6]   Beagle: Automated Extraction and Interpretation of Visualizations from the Web [J].
Battle, Leilani ;
Duan, Peitong ;
Miranda, Zachery ;
Mukusheva, Dana ;
Chang, Remco ;
Stonebraker, Michael .
PROCEEDINGS OF THE 2018 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI 2018), 2018,
[7]   D3: Data-Driven Documents [J].
Bostock, Michael ;
Ogievetsky, Vadim ;
Heer, Jeffrey .
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, 2011, 17 (12) :2301-2309
[8]  
Brown TB, 2020, ADV NEUR IN, V33
[9]  
Card S.K., 1999, Readings in Information Visualization: Using Vision to Think
[10]   Honeybee: Locality-enhanced Projector for Multimodal LLM [J].
Cha, Junbum ;
Kang, Wooyoung ;
Mun, Jonghwan ;
Roh, Byungseok .
2024 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2024, :13817-13827