DNN推理早期就基本确定输出，偏见是关键驱动因素
DNNs May Determine Major Properties of Their Outputs Early, with Timing Possibly Driven by Bias
- 模型早期阶段已决定输出，由内部偏见主导决策
- 扩散模型实验显示设计与训练中的偏见影响早期判断
- 适合关注模型可解释性与高效推理的研究者
本文提出，深度神经网络（DNN）在推理的早期阶段即基本确定输出，模型固有的偏见在这一过程中起关键作用。我们以扩散模型（DMs）为案例，表明DNN的早期决策受其设计和训练中偏见类型与程度的影响。该发现为偏见缓解、高效推理及机器学习系统解释提供了新视角。通过揭示DNN决策的时间动态，本文旨在激发机器学习领域更深入的讨论与研究。
原文摘要 · Abstract (English)
This paper argues that deep neural networks (DNNs) mostly determine their outputs during the early stages of inference, where biases inherent in the model play a crucial role in shaping this process. We draw a parallel between this phenomenon and human decision-making, which often relies on fast, intuitive heuristics. Using diffusion models (DMs) as a case study, we demonstrate that DNNs often make early-stage decision-making influenced by the type and extent of bias in their design and training. Our findings offer a new perspective on bias mitigation, efficient inference, and the interpretation of machine learning systems. By identifying the temporal dynamics of decision-making in DNNs, this paper aims to inspire further discussion and research within the machine learning community.
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。