# PINN-Learnable-Activation

**Repository Path**: hbwei/PINN-Learnable-Activation

## Basic Information

- **Project Name**: PINN-Learnable-Activation
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-05-01
- **Last Updated**: 2026-05-01

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# PINN-Learnable-Activation
Exploring the impact of Learnable Activation functions on PINNs

This is the codebase for my course project for MAT1510 - Deep Learning: Theory and Data Science. 

In this course project, I build on the work of the PINNsformer (https://arxiv.org/pdf/2307.11833) and their Wavelet Activation function to understand the impact of parameterized, trainable activation functions on other Physics-Inspired deep learning approaches such as standard PINNs and Quadratic Residual Networks, and on the loss landscape being traversed.

As part of this project I explore the following activation functions: Sine Activation, Hyperbolic Tangent, Wavelet (from PINNsformer), Periodic Wavelet (Wavelet with period parameters), LearnableLeakyRelu, Parameterized Swish, and Logmoid Activation Unit (LAU).