The Role of Tuning Uncertain Inference Systems

TL;DR

This paper investigates the impact of tuning parameters in various uncertain inference models, revealing that simpler models like the independence model can outperform more complex ones like MYCIN and PROSPECTOR when optimized.

Contribution

It demonstrates that tuning parameters in different inference systems can lead to comparable or improved accuracy, challenging assumptions about model complexity.

Findings

Independence model outperformed MYCIN and PROSPECTOR after tuning.

All models performed similarly when optimally tuned.

Simpler models can be more accurate than complex ones in certain settings.

Abstract

This study examined the effects of "tuning" the parameters of the incremental function of MYCIN, the independent function of PROSPECTOR, a probability model that assumes independence, and a simple additive linear equation. me parameters of each of these models were optimized to provide solutions which most nearly approximated those from a full probability model for a large set of simple networks. Surprisingly, MYCIN, PROSPECTOR, and the linear equation performed equivalently; the independence model was clearly more accurate on the networks studied.