Understanding and Mitigating Memorization in Diffusion Models for Tabular Data

Guardado en:
書目詳細資料
發表在:arXiv.org (Dec 15, 2024), p. n/a
主要作者: Fang, Zhengyu
其他作者: Jiang, Zhimeng, Chen, Huiyuan, Li, Xiao, Li, Jing
出版:
Cornell University Library, arXiv.org
主題:
在線閱讀:Citation/Abstract
Full text outside of ProQuest
標簽: 添加標簽
沒有標簽, 成為第一個標記此記錄!

MARC

LEADER 00000nab a2200000uu 4500
001 3145907151
003 UK-CbPIL
022 |a 2331-8422 
035 |a 3145907151 
045 0 |b d20241215 
100 1 |a Fang, Zhengyu 
245 1 |a Understanding and Mitigating Memorization in Diffusion Models for Tabular Data 
260 |b Cornell University Library, arXiv.org  |c Dec 15, 2024 
513 |a Working Paper 
520 3 |a Tabular data generation has attracted significant research interest in recent years, with the tabular diffusion models greatly improving the quality of synthetic data. However, while memorization, where models inadvertently replicate exact or near-identical training data, has been thoroughly investigated in image and text generation, its effects on tabular data remain largely unexplored. In this paper, we conduct the first comprehensive investigation of memorization phenomena in diffusion models for tabular data. Our empirical analysis reveals that memorization appears in tabular diffusion models and increases with larger training epochs. We further examine the influence of factors such as dataset sizes, feature dimensions, and different diffusion models on memorization. Additionally, we provide a theoretical explanation for why memorization occurs in tabular diffusion models. To address this issue, we propose TabCutMix, a simple yet effective data augmentation technique that exchanges randomly selected feature segments between random same-class training sample pairs. Building upon this, we introduce TabCutMixPlus, an enhanced method that clusters features based on feature correlations and ensures that features within the same cluster are exchanged together during augmentation. This clustering mechanism mitigates out-of-distribution (OOD) generation issues by maintaining feature coherence. Experimental results across various datasets and diffusion models demonstrate that TabCutMix effectively mitigates memorization while maintaining high-quality data generation. 
653 |a Datasets 
653 |a Data augmentation 
653 |a Tables (data) 
653 |a Empirical analysis 
653 |a Image quality 
653 |a Clustering 
653 |a Synthetic data 
700 1 |a Jiang, Zhimeng 
700 1 |a Chen, Huiyuan 
700 1 |a Li, Xiao 
700 1 |a Li, Jing 
773 0 |t arXiv.org  |g (Dec 15, 2024), p. n/a 
786 0 |d ProQuest  |t Engineering Database 
856 4 1 |3 Citation/Abstract  |u https://www.proquest.com/docview/3145907151/abstract/embedded/ZKJTFFSVAI7CB62C?source=fedsrch 
856 4 0 |3 Full text outside of ProQuest  |u http://arxiv.org/abs/2412.11044