Skip to content

Repository for "Inverse Design of Silicon Photonics Components: A Study from Deep Learning Perspective"

Notifications You must be signed in to change notification settings

dreitan1/IPPSDesign

Repository files navigation

IPPSDesign

Repository for "Inverse Design of Silicon Photonics Components: A Study from Deep Learning Perspective"

Results

Whole-data evaluation

results

Our results show that our ResNet and InceptionNet models both perform reasonably well for PPS generation, especially when compared with VGG. Our results when trained on 90% of the data and tested on the remaining 10% are concurrent withthe results found in our whole-data evaluation.

Requirements

Load requirements with:

python3 -m pip install -r requirements.txt

Training the Forward Model

python3 -m train_forward [--path=PATH]

PATH as default is ".."

Training the Inverse Model

python3 -m train_inverse --model=MODEL [--model_name=NAME] [--path=PATH]

MODEL can be ResNet, InceptionNet, or VGG

NAME as default is "res_model.pth", "inception_model.pth", or "vgg_model.pth", depending on MODEL

PATH as default is ".."

If model name is not specified, defaults to name recognized by default tester, based on model used.

Input structures and output characteristics expected to be in folder at $path + "Machine learning inverse design/input_patterns" and $path + "Machine learning inverse design/output_characteristics" respectively.

This code trains on the entire dataset. To instead run the training with a 90:10 train-test split, replace train_inverse with train_inverse9010

Testing the Models

python3 -m test_inverse --model=MODEL [--model_name=NAME] [--path=PATH]

MODEL can be ResNet, InceptionNet, or VGG

NAME as default is "res_model.pth", "inception_model.pth", or "vgg_model.pth", depending on MODEL

PATH as default is ".."

To run this program, two models are required in the models directory: the inverse model to test and the forward model (forward_model.pth).

Misc.

All models save to/load from the '/models' directory, located at the specified path. All figures save to the '/figures' directory, also located at that path.

Pre-trained models and data can be downloaded from: https://drive.google.com/drive/folders/1QRARk_RdEI-WgjzwzjKGTikWUeJLO2DH?usp=sharing

About

Repository for "Inverse Design of Silicon Photonics Components: A Study from Deep Learning Perspective"

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages