Loading Now

Summary of Polynomial-augmented Neural Networks (panns) with Weak Orthogonality Constraints For Enhanced Function and Pde Approximation, by Madison Cooley et al.


Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation

by Madison Cooley, Shandian Zhe, Robert M. Kirby, Varun Shankar

First submitted to arxiv on: 4 Jun 2024

Categories

  • Main: Machine Learning (cs.LG)
  • Secondary: None

     Abstract of paper      PDF of paper


GrooveSquid.com Paper Summaries

GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!

Summary difficulty Written by Summary
High Paper authors High Difficulty Summary
Read the original abstract here
Medium GrooveSquid.com (original content) Medium Difficulty Summary
We introduce Polynomial-Augmented Neural Networks (PANNs), a novel architecture combining deep neural networks (DNNs) and polynomial approximation. PANNs leverage the strengths of both components, offering flexibility and efficiency in higher-dimensional approximation, along with rapid convergence rates for smooth functions. To ensure stable training and enhanced accuracy, we propose orthogonality constraints, basis pruning, and polynomial preconditioning strategies. We experimentally demonstrate PANNs’ superior approximation properties over DNNs for regression and partial differential equations (PDEs) solution, as well as improved accuracy over both polynomial and DNN-based regression when regressing functions with limited smoothness.
Low GrooveSquid.com (original content) Low Difficulty Summary
We created a new way to make neural networks better at solving problems. We combined two ideas: deep neural networks (good for complex tasks) and polynomials (fast and simple). Our new architecture, called Polynomial-Augmented Neural Networks (PANNs), does this combination in a clever way. This helps it solve some problems more accurately than just using one or the other alone. We tested PANNs on different types of problems and showed that it’s better at solving certain kinds of equations and functions.

Keywords

» Artificial intelligence  » Pruning  » Regression