用新指标量化数据噪声对模型性能的影响
Towards Modeling Data Quality and Machine Learning Model Performance
- 提出DDR指标衡量数据不确定性与模型性能关系
- 实验显示准确率随DDR变化，可画出性能曲线
- 适合关注数据质量与模型可信度的研究者
理解数据中的不确定性和噪声对机器学习模型（MLM）的影响，对于建立信任和评估性能至关重要。本文提出一种新模型，用于量化数据中的不确定性与噪声对MLM的影响。基于信噪比（SNR）概念，引入新的指标——确定性-非确定性比（DDR），以描述模型性能。通过合成数据实验，我们展示了准确率如何随DDR变化，并利用DDR-准确率曲线来评估模型性能。
原文摘要 · Abstract (English)
Understanding the effect of uncertainty and noise in data on machine learning models (MLM) is crucial in developing trust and measuring performance. In this paper, a new model is proposed to quantify uncertainties and noise in data on MLMs. Using the concept of signal-to-noise ratio (SNR), a new metric called deterministic-non-deterministic ratio (DDR) is proposed to formulate performance of a model. Using synthetic data in experiments, we show how accuracy can change with DDR and how we can use DDR-accuracy curves to determine performance of a model.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。