# Exploring the Performance of Perforated Backpropagation through Further Experiments

**Authors:** Rorry Brenner, Evan Davis, Rushi Chaudhari, Rowan Morse, Jingyao Chen, Xirui Liu, Zhaoyi You, Laurent Itti  
**arXiv:** [2506.00356](https://arxiv.org/abs/2506.00356)  
**Date:** 2025-05-31  
**Summary page:** https://www.summarizepaper.com/en/paper/2506.00356v1/

## Key points

- Cutting-edge neural network optimization technique
- Leverages the computational significance of dendrites in biological neurons
- Hackathon in February 2025
- Participants included students and local machine learning practitioners from Pittsburgh
- Focus on exploring Perforated Backpropagation algorithm and its application on datasets and models
- Up to a staggering 90% reduction in model size without compromising accuracy
- Some projects saw up to a 16% boost in accuracy compared to original models
- Rorry Brenner, Evan Davis, Rushi Chaudhari, Rowan Morse, Jingyao Chen, Xirui Liu, Zhaoyi You, Laurent Itti
- Promising implications of Perforated Backpropagation in advancing neural network optimization techniques
- Provides valuable insights for computer science and artificial intelligence fields

## Summary

Perforated Backpropagation is a cutting-edge neural network optimization technique that leverages the computational significance of dendrites in biological neurons. This innovative approach was further explored through additional experiments conducted at a hackathon hosted at the Carnegie Mellon Swartz Center in February 2025. The event brought together students and local machine learning practitioners from Pittsburgh to delve into the Perforated Backpropagation algorithm and its application on their respective datasets and models. The results of these experiments were nothing short of remarkable, showcasing the potential for significant enhancements in project outcomes. Notably, participants observed up to a staggering 90% reduction in model size without compromising accuracy. Additionally, some projects experienced an impressive boost of up to 16% in accuracy compared to their original models. The study, authored by Rorry Brenner, Evan Davis, Rushi Chaudhari, Rowan Morse, Jingyao Chen, Xirui Liu, Zhaoyi You, and Laurent Itti, sheds light on the promising implications of Perforated Backpropagation in advancing neural network optimization techniques. With 10 pages of detailed analysis accompanied by 7 figures and 1 table for reference,this research contributes valuable insights to the fields of computer science and artificial intelligence.