arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.16253cs.LG
收藏

学习去帮助:训练模型以协助遗留设备

Learning To Help: Training Models to Assist Legacy Devices

Rutgers, The State University of New Jersey(罗格斯大学(新泽西州立大学))

Yu Wu, Anand Sarwate

AI总结:

针对遗留设备计算能力受限问题,本文将其置于带弃权学习框架中,形式化了为固定客户端训练专家的逆向问题,推导出贝叶斯最优规则与一致替代损失函数,实验证明该方法优于基于置信度的拒绝规则。

摘要:

在物理设备硬件上实现的机器学习模型可能会被部署很长时间。设备的计算能力可能有限,并且相对于较新的改进会变得过时。由于机器学习模型的规模,卸载一些计算(例如卸载到边缘云)可以帮助此类遗留设备。我们将此问题置于带弃权(不执行)学习(LWA)的框架中,在该框架中必须训练专家(边缘)以协助客户端(设备)。先前关于LWA的工作在假设边缘要么是预言机要么是人类专家的情况下训练客户端。在这项工作中,我们形式化了为固定(遗留)客户端训练专家的逆向问题。与LWA中一样,客户端使用拒绝规则来决定何时将推理卸载给专家(需要付出成本)。我们找到了贝叶斯最优规则,证明了一个泛化界,并找到了一致的替代损失函数。实验结果表明,我们的框架优于基于置信度的拒绝规则。

英文摘要:

Machine learning models implemented in hardware on physical devices may be deployed for a long time. The computational abilities of the device may be limited and become outdated with respect to newer improvements. Because of the size of ML models, offloading some computation (e.g. to an edge cloud) can help such legacy devices. We cast this problem in the framework of learning with abstention (LWA) in which the expert (edge) must be trained to assist the client (device). Prior work on LWA trains the client assuming the edge is either an oracle or a human expert. In this work, we formalize the reverse problem of training the expert for a fixed (legacy) client. As in LWA, the client uses a rejection rule to decide when to offload inference to the expert (at a cost). We find the Bayes-optimal rule, prove a generalization bound, and find a consistent surrogate loss function. Empirical results show that our framework outperforms confidence-based rejection rules.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑