Surrogate- and Invariance-Boosted Contrastive Learning (SIB-CL)

Related tags

Deep LearningSIB-CL
Overview

Surrogate- and Invariance-Boosted Contrastive Learning (SIB-CL)

This repository contains all source code used to generate the results in the article "Surrogate- and invariance-boosted contrastive learning for data-scarce applications in science". (url: to-be-updated)

  • The folder generate_datasets contains all numerical programs used to generate the datasets, for both Photonic Crystals (PhC) and the Time-independent Schrodinger Equation (TISE)
  • main.py is the main code used to train the neural networks (explained in detail below)

Dependencies

Please install the required Python packages: pip install -r requirements.txt

A python3 environment can be created prior to this: conda create -n sibcl python=3.8; conda activate sibcl

Assess to MATLAB is required to calculate the density-of-states (DOS) of PhCs.

Dataset Generation

Photonic Crystals (PhCs)

Relevant code stored in generate_datasets/PhC/. Periodic unit cells are defined using a level set of a Fourier sum; different unit cells can be generated using the get_random() method of the FourierPhC class defined in fourier_phc.py.

To generate the labeled PhC datasets, we first compute their band structures using MPB. This can be executed via:

For the target dataset of random fourier unit cells, python phc_gendata.py --h5filename="mf1-s1" --pol="tm" --nsam=5000 --maxF=1 --seed=1;

and for the source dataset of simple cylinders, python phc_gencylin.py --h5filename="cylin" --pol="tm" --nsam=10000;

each program will create a dataset with the eigen-frequencies, group velocities, etc, stored in a .h5 file (which can be accessed using the h5py package). We then calculate the DOS using the GRR method provided by the MATLAB code https://github.com/boyuanliuoptics/DOS-calculation/blob/master/DOS_GGR.m. To do so, we first parse the data to create the .txt files required as inputs to the program, compute the DOS using MATLAB and then add the DOS labels back to the original .h5 files. These steps will be executed automatically by simply running the shell script get_DOS.sh after modifying the h5 filename identifier defined at the top. Note that for this to run smoothly, python and MATLAB will first need to be added to PATH.

Time-independent Schrodinger Equation (TISE)

Relevant code stored in generate_datasets/TISE/. Example usage:

To generate target dataset, e.g. in 3D, python tise_gendata.py --h5filename="tise3d" --ndim 3 --nsam 5000

To generate low resolution dataset, python tise_gendata.py --h5filename='tise3d_lr' --ndim 3 --nsam 10000 --lowres --orires=32 (--orires defines the resolution of the input to the neural network)

To generate qho dataset, python tise_genqho.py --h5filename='tise2d_qho' --ndim 2 --nsam 10000

SIB-CL and baselines training

Training of the neural networks for all problems introduced in the article (i.e. PhC DOS prediction, PhC Band structure prediction, TISE ground state energy prediction using both low resolution or QHO data as surrogate) can all be executed using main.py by indicating the appropriate flags (see below). This code also allows training via the SIB-CL framework or any of the baselines, again with the use of the appropriate flag. This code also contains other prediction problems not presented in the article, such as predicting higher energy states of TISE, TISE wavefunctions and single band structure.

Important flags:

--path_to_h5: indicate directory where h5 datasets are located. The h5 filenames defined in the dataset classes in datasets_PhC_SE.py should also be modified according to the names used during dataset generation.

--predict: defines prediction task. Options: 'DOS', 'bandstructures', 'eigval', 'oneband', 'eigvec'

--train: specify if training via SIB-CL or baselines. Options: 'sibcl', 'tl', 'sl', 'ssl' ('ssl' performs regular contrastive learning without surrogate dataset). For invariance-boosted baselines, e.g. TL-I or SL-I, specify 'tl' or 'sl' here and add the relevant invariances flags (see below).

--iden: required; specify identifier for saving of models, training logs and results

Invariances flags: --translate_pbc (set this flag to include rolling translations), --pg_uniform (set this flag to uniformly sample the point group symmetry transformations), --scale (set this flag to scale unit cell - used for PhC), --rotate (set this flag to do 4-fold rotations), --flip (set this flag to perform horizontal and vertical mirrors). If --pg_uniform is used, there is no need to include --rotate and --flip.

Other optional flags can be displayed via python main.py --help. Examples of shell scripts can be found in the sh_scripts folder.

Training outputs:

By default, running main.py will create 3 subdirectories:

  • ./pretrained_models/: state dictionaries of pretrained models at various epochs indicated in the eplist variable will be saved to this directory. These models are used for further fine-tuning.
  • ./dicts/: stores the evaluation losses on the test set as dictionaries saved as .json files. The results can then be plotted using plot_results.py.
  • ./tlogs/: training curves for pre-training and fine-tuning are stored in dictionaries saved as .json files. The training curves can be plotted using get_training_logs.py. Alternatively, the --log_to_tensorboard flag can be set and training curves can be viewed using tensorboard; in this case, the dictionaries will not be generated.
You might also like...
pytorch implementation of
pytorch implementation of "Contrastive Multiview Coding", "Momentum Contrast for Unsupervised Visual Representation Learning", and "Unsupervised Feature Learning via Non-Parametric Instance-level Discrimination"

Unofficial implementation: MoCo: Momentum Contrast for Unsupervised Visual Representation Learning (Paper) InsDis: Unsupervised Feature Learning via N

Dense Contrastive Learning (DenseCL) for self-supervised representation learning, CVPR 2021.
Dense Contrastive Learning (DenseCL) for self-supervised representation learning, CVPR 2021.

Dense Contrastive Learning for Self-Supervised Visual Pre-Training This project hosts the code for implementing the DenseCL algorithm for se

CRLT: A Unified Contrastive Learning Toolkit for Unsupervised Text Representation Learning
CRLT: A Unified Contrastive Learning Toolkit for Unsupervised Text Representation Learning

CRLT: A Unified Contrastive Learning Toolkit for Unsupervised Text Representation Learning This repository contains the code and relevant instructions

Source code and dataset for ACL2021 paper: "ERICA: Improving Entity and Relation Understanding for Pre-trained Language Models via Contrastive Learning".

ERICA Source code and dataset for ACL2021 paper: "ERICA: Improving Entity and Relation Understanding for Pre-trained Language Models via Contrastive L

PyTorch implementation of
PyTorch implementation of "Supervised Contrastive Learning" (and SimCLR incidentally)

PyTorch implementation of "Supervised Contrastive Learning" (and SimCLR incidentally)

VIMPAC: Video Pre-Training via Masked Token Prediction and Contrastive Learning

This is a release of our VIMPAC paper to illustrate the implementations. The pretrained checkpoints and scripts will be soon open-sourced in HuggingFace transformers.

Code for 'Single Image 3D Shape Retrieval via Cross-Modal Instance and Category Contrastive Learning', ICCV 2021
Code for 'Single Image 3D Shape Retrieval via Cross-Modal Instance and Category Contrastive Learning', ICCV 2021

CMIC-Retrieval Code for Single Image 3D Shape Retrieval via Cross-Modal Instance and Category Contrastive Learning. ICCV 2021. Introduction In this wo

Official Pytorch implementation of "Unbiased Classification Through Bias-Contrastive and Bias-Balanced Learning (NeurIPS 2021)

Unbiased Classification Through Bias-Contrastive and Bias-Balanced Learning (NeurIPS 2021) Official Pytorch implementation of Unbiased Classification

This is the repository for the AAAI 21 paper [Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning].

CG3 This is the repository for the AAAI 21 paper [Contrastive and Generative Graph Convolutional Networks for Graph-based Semi-Supervised Learning]. R

Releases(v1.0)
Owner
Charlotte Loh
PhD candidate at MIT EECS
Charlotte Loh
Full Stack Deep Learning Labs

Full Stack Deep Learning Labs Welcome! Project developed during lab sessions of the Full Stack Deep Learning Bootcamp. We will build a handwriting rec

Full Stack Deep Learning 1.2k Dec 31, 2022
CLDF dataset derived from Robbeets et al.'s "Triangulation Supports Agricultural Spread" from 2021

CLDF dataset derived from Robbeets et al.'s "Triangulation Supports Agricultural Spread" from 2021 How to cite If you use these data please cite the o

Digital Linguistics 2 Dec 20, 2021
This is an official implementation for "Video Swin Transformers".

Video Swin Transformer By Ze Liu*, Jia Ning*, Yue Cao, Yixuan Wei, Zheng Zhang, Stephen Lin and Han Hu. This repo is the official implementation of "V

Swin Transformer 981 Jan 03, 2023
Open Source Light Field Toolbox for Super-Resolution

BasicLFSR BasicLFSR is an open-source and easy-to-use Light Field (LF) image Super-Ressolution (SR) toolbox based on PyTorch, including a collection o

Squidward 50 Nov 18, 2022
《Train in Germany, Test in The USA: Making 3D Object Detectors Generalize》(CVPR 2020)

Train in Germany, Test in The USA: Making 3D Object Detectors Generalize This paper has been accpeted by Conference on Computer Vision and Pattern Rec

Xiangyu Chen 101 Jan 02, 2023
Pytorch implementation of "M-LSD: Towards Light-weight and Real-time Line Segment Detection"

M-LSD: Towards Light-weight and Real-time Line Segment Detection Pytorch implementation of "M-LSD: Towards Light-weight and Real-time Line Segment Det

123 Jan 04, 2023
这个开源项目主要是对经典的时间序列预测算法论文进行复现,模型主要参考自GluonTS,框架主要参考自Informer

Time Series Research with Torch 这个开源项目主要是对经典的时间序列预测算法论文进行复现,模型主要参考自GluonTS,框架主要参考自Informer。 建立原因 相较于mxnet和TF,Torch框架中的神经网络层需要提前指定输入维度: # 建立线性层 TensorF

Chi Zhang 85 Dec 29, 2022
[ICCV 2021] Focal Frequency Loss for Image Reconstruction and Synthesis

Focal Frequency Loss - Official PyTorch Implementation This repository provides the official PyTorch implementation for the following paper: Focal Fre

Liming Jiang 460 Jan 04, 2023
Exploring the link between uncertainty estimates obtained via "exact" Bayesian inference and out-of-distribution (OOD) detection.

Uncertainty-based OOD detection Exploring the link between uncertainty estimates obtained by "exact" Bayesian inference and out-of-distribution (OOD)

Christian Henning 1 Nov 05, 2022
A containerized REST API around OpenAI's CLIP model.

OpenAI's CLIP — REST API This is a container wrapping OpenAI's CLIP model in a RESTful interface. Running the container locally First, build the conta

Santiago Valdarrama 48 Nov 06, 2022
Official repository for the ICLR 2021 paper Evaluating the Disentanglement of Deep Generative Models with Manifold Topology

Official repository for the ICLR 2021 paper Evaluating the Disentanglement of Deep Generative Models with Manifold Topology Sharon Zhou, Eric Zelikman

Stanford Machine Learning Group 34 Nov 16, 2022
Evaluation and Benchmarking of Speech Super-resolution Methods

Speech Super-resolution Evaluation and Benchmarking What this repo do: A toolbox for the evaluation of speech super-resolution algorithms. Unify the e

Haohe Liu (刘濠赫) 84 Dec 20, 2022
Source code for "Progressive Transformers for End-to-End Sign Language Production" (ECCV 2020)

Progressive Transformers for End-to-End Sign Language Production Source code for "Progressive Transformers for End-to-End Sign Language Production" (B

58 Dec 21, 2022
6D Grasping Policy for Point Clouds

GA-DDPG [website, paper] Installation git clone https://github.com/liruiw/GA-DDPG.git --recursive Setup: Ubuntu 16.04 or above, CUDA 10.0 or above, py

Lirui Wang 48 Dec 21, 2022
RuDOLPH: One Hyper-Modal Transformer can be creative as DALL-E and smart as CLIP

[Paper] [Хабр] [Model Card] [Colab] [Kaggle] RuDOLPH 🦌 🎄 ☃️ One Hyper-Modal Tr

Sber AI 230 Dec 31, 2022
An unofficial PyTorch implementation of a federated learning algorithm, FedAvg.

Federated Averaging (FedAvg) in PyTorch An unofficial implementation of FederatedAveraging (or FedAvg) algorithm proposed in the paper Communication-E

Seok-Ju Hahn 123 Jan 06, 2023
Feature board for ERPNext

ERPNext Feature Board Feature board for ERPNext Development Prerequisites k3d kubectl helm bench Install K3d Cluster # export K3D_FIX_CGROUPV2=1 # use

Revant Nandgaonkar 16 Nov 09, 2022
text_recognition_toolbox: The reimplementation of a series of classical scene text recognition papers with Pytorch in a uniform way.

text recognition toolbox 1. 项目介绍 该项目是基于pytorch深度学习框架,以统一的改写方式实现了以下6篇经典的文字识别论文,论文的详情如下。该项目会持续进行更新,欢迎大家提出问题以及对代码进行贡献。 模型 论文标题 发表年份 模型方法划分 CRNN 《An End-t

168 Dec 24, 2022
Code for paper: Group-CAM: Group Score-Weighted Visual Explanations for Deep Convolutional Networks

Group-CAM By Zhang, Qinglong and Rao, Lu and Yang, Yubin [State Key Laboratory for Novel Software Technology at Nanjing University] This repo is the o

zhql 98 Nov 16, 2022
The repo for the paper "I3CL: Intra- and Inter-Instance Collaborative Learning for Arbitrary-shaped Scene Text Detection".

I3CL: Intra- and Inter-Instance Collaborative Learning for Arbitrary-shaped Scene Text Detection Updates | Introduction | Results | Usage | Citation |

33 Jan 05, 2023