# TropNNC: Structured Neural Network Compression Using Tropical Geometry

Source: [TropNNC: Structured Neural Network Compression Using Tropical Geometry](<https://www.sota2.com/research/paper/68fa75a11e209131cea69129>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-12-25T05:00:04.588Z

Authors: Konstantinos Fotopoulos, Petros Maragos, Panagiotis Misiakos

Published: 2024-09-05T23:54:32.000Z

## Abstract

We present TropNNC, a framework for compressing neural networks with linear and convolutional layers and ReLU activations using tropical geometry. By representing a network's output as a tropical rational function, TropNNC enables structured compression via reduction of the corresponding tropical polynomials. Our method refines the geometric approximation of previous work by adaptively selecting the weights of retained neurons. Key contributions include the first application of tropical geometry to convolutional layers and the tightest known theoretical compression bound. TropNNC requires only access to network weights - no training data - and achieves competitive performance on MNIST, CIFAR, and ImageNet, matching strong baselines such as ThiNet and CUP.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2409.03945>)

[PDF](<https://arxiv.org/pdf/2409.03945.pdf>)