# Revisiting Multi-Step Nonlinearity Compensation with Machine Learning

## TL;DR

This paper challenges the conventional wisdom that fewer steps are always better for fiber nonlinearity compensation, demonstrating that multi-step approaches can achieve superior performance with balanced complexity.

## Contribution

It introduces a novel perspective that carefully designed multi-step methods outperform simpler approaches in fiber nonlinearity compensation.

## Key findings

- Multi-step approaches can outperform fewer-step methods.
- Careful design of multi-step processes improves performance.
- Trade-offs between complexity and effectiveness are better balanced with multi-step methods.

## Abstract

For the efficient compensation of fiber nonlinearity, one of the guiding principles appears to be: fewer steps are better and more efficient. We challenge this assumption and show that carefully designed multi-step approaches can lead to better performance-complexity trade-offs than their few-step counterparts.