Promoting Simple Agents: Ensemble Methods for Event-Log Prediction

TL;DR

This paper compares lightweight n-gram models with neural architectures for event-log prediction, showing that n-grams can achieve similar accuracy with less resources and proposing a dynamic ensemble method to optimize performance.

Contribution

It introduces the promotion ensemble algorithm that dynamically selects models during inference, reducing computational overhead while maintaining high accuracy.

Findings

N-grams achieve comparable accuracy to neural models with fewer resources.

The promotion ensemble method reduces inference overhead compared to classical voting.

Ensembles with the promotion algorithm match or surpass neural models on real datasets.

Abstract

We compare lightweight automata-based models (n-grams) with neural architectures (LSTM, Transformer) for next-activity prediction in streaming event logs. Experiments on synthetic patterns and five real-world process mining datasets show that n-grams with appropriate context windows achieve comparable accuracy to neural models while requiring substantially fewer resources. Unlike windowed neural architectures, which show unstable performance patterns, n-grams provide stable and consistent accuracy. While we demonstrate that classical ensemble methods like voting improve n-gram performance, they require running many agents in parallel during inference, increasing memory consumption and latency. We propose an ensemble method, the promotion algorithm, that dynamically selects between two active models during inference, reducing overhead compared to classical voting schemes. On real-world datasets, these ensembles match or exceed the accuracy of non-windowed neural models with lower computational cost.