Code for "SUGAR: Subgraph Neural Network with Reinforcement Pooling and Self-Supervised Mutual Information Mechanism"

Related tags

Testing SUGAR
Overview

SUGAR

Code for "SUGAR: Subgraph Neural Network with Reinforcement Pooling and Self-Supervised Mutual Information Mechanism"

Overview

  • train.py: the core of our model, including the structure and the process of training.
  • env.py, QLearning.py: the code about RL method
  • GCN.py, layers.py: including the basic layers we used in the main model.
  • dataset/: including the dataset:MUTAG, DD, NCI1, NCI109, PTC_MR, ENZYMES, PROTEINS.
    • 'RAW/': the original data of the dataset
    • adj.npy: the biggest Adjacency Matrix built from dataset
    • graph_label.npy: the label of every sub_graph
    • sub_adj.npy: the Adjacency Matrix of subgraph through sampling
    • features.npy: the pre-processed features of each subgraph

Datasets

  • MUTAG: The MUTAG dataset consists of 188 chemical compounds divided into two classes according to their mutagenic effect on a bacterium.
  • D&D: D&D is a dataset of 1178 protein structures (Dobson and Doig, 2003). Each protein is represented by a graph, in which the nodes are amino acids and two nodes are connected by an edge if they are less than 6 Angstroms apart. The prediction task is to classify the protein structures into enzymes and non-enzymes.
  • NCI1&NCI109:NCI1 and NCI109 represent two balanced subsets of datasets of chemical compounds screened for activity against non-small cell lung cancer and ovarian cancer cell lines respectively (Wale and Karypis (2006) and http://pubchem.ncbi.nlm.nih.gov).
  • ENZYMES: ENZYMES is a dataset of protein tertiary structures obtained from (Borgwardt et al., 2005) consisting of 600 enzymes from the BRENDA enzyme database (Schomburg et al., 2004). In this case the task is to correctly assign each enzyme to one of the 6 EC top-level classes.

Setting

  1. setting python env using pip install -r requirements.txt
  2. cd ./dataset &python transform.py --dataset MUTAG
  3. python train.py(all the parameters could be viewed in the train.py)

Parameters

     --dataset DATASET
     --num_info NUM_INFO
     --lr LR (learning_rate)
     --max_pool MAX_POOL
     --momentum MOMENTUM
     --num_epoch NUM_EPOCH
     --batch_size BATCH_SIZE
     --sg_encoder SG_ENCODER(GIN, GCN, GAT, SAGE)
     --MI_loss MI_LOSS
     --start_k START_K

Reference

@inproceedings{sun2021sugar,
  title={SUGAR: Subgraph Neural Network with Reinforcement Pooling and Self-Supervised Mutual Information Mechanism},
  author={Sun, Qingyun and Li, Jianxin and Peng, Hao and Wu, Jia and Ning, Yuanxing and Yu, Phillip S and He, Lifang},
  booktitle={Proceedings of the 2021 World Wide Web Conference},
  year={2021}
}
Comments
  • hyperparameter settings request,or training log

    hyperparameter settings request,or training log

    Thanks for answering my last question,but our team still didn't get the performance in the paper.Can we get the hyperparameter settings in MuTag,NCI1,PTC_MR,Protein.or your training logs in these datasets?

    opened by HuangQY123 4
  • Is there any ablation about the two losses?

    Is there any ablation about the two losses?

    Hi @Suchun-sv @SunQingYun1996,

    Thanks for the great work! I wonder if there are any ablation studies such as: SUGAR w/o supervised loss or SUGAR w/o self-supervised contrastive loss or SUGAR w/o reinforcement pooling. I think it will validate the effects of each module better. I'm studying unsupervised graph representation learning and the effect of SUGAR w/o supervised loss could inspire me in some way.

    Thanks!

    opened by ha-lins 4
  • Tensorflow version problems

    Tensorflow version problems

    Hello, thanks for your nice work on graph classification. Recently, when I use your code to run experiments, I use pip install -r requirement.txt, but pip tells me that tensorflow1.15 requires tensorboard>1.15 <1.16, but tensorflow-gpu2.3 needs tensorboard>2.3. So I am eager to know how to build a env with correct tf version to run your code, thanks.

    opened by lunaticbg 2
  • Gradient explosion and memory error in DD,NCI1,NCI109

    Gradient explosion and memory error in DD,NCI1,NCI109

    1. In DD: Traceback (most recent call last): File "transform.py", line 443, in adjs, d_es = paser.main(save=True) File "transform.py", line 153, in main d_es, adj_com = self.ex_edges() File "transform.py", line 87, in ex_edges adj = np.zeros((self.n, self.n)) numpy.core._exceptions.MemoryError: Unable to allocate 112. GiB for an array with shape (122494, 122494) and data type float64 2.In NCI1 and NCI109: folds 1/10: 0%| | 1/1000 [04:11<41:05:52, 148.10s/it, k:0.80, loss: nan, best_acc:1.00, RL:0]
      folds 1/10: 0%| | 2/1000 [04:11<33:48:52, 121.98s/it, k:0.80, loss: nan, best_acc:1.00, RL:0] folds 1/10: 0%| | 2/1000 [05:52<33:48:52, 121.98s/it, k:0.80, loss: nan, best_acc:1.00, RL:0] folds 1/10: 0%| | 2/1000 [05:52<33:48:52, 121.98s/it, k:0.80, loss: nan, best_acc:1.00, RL:0] folds 1/10: 0%| | 3/1000 [05:52<31:02:52, 112.11s/it, k:0.80, loss: nan, best_acc:1.00, RL:0]
    opened by HuangQY123 2
  • Exploding gradients problem and binary DD release

    Exploding gradients problem and binary DD release

    1.Could you please give me the binary DD output and release again? The link is dead。 2.I also encountered the problem of exploding gradients, could you please give the details of the training parameters for the NCI1 or NCI109 dataset? Such as learning rate, batch size, hidden layer settings, etc. I shall be profoundly grateful if you will give me an answer.

    opened by id3509 1
Owner
null
GitHub
py.test fixture for benchmarking code

Overview docs tests package A pytest fixture for benchmarking code. It will group the tests into rounds that are calibrated to the chosen timer. See c

Ionel Cristian Mărieș 1k Jan 3, 2023
To automate the generation and validation tests of COSE/CBOR Codes and it's base45/2D Code representations

To automate the generation and validation tests of COSE/CBOR Codes and it's base45/2D Code representations, a lot of data has to be collected to ensure the variance of the tests. This respository was established to collect a lot of different test data and related test cases of different member states in a standardized manner. Each member state can generate a folder in this section.

null 160 Jul 25, 2022
Rerun pytest when your code changes

A simple watcher for pytest Overview pytest-watcher is a tool to automatically rerun pytest when your code changes. It looks for the following events:

Olzhas Arystanov 74 Dec 29, 2022
Obsei is a low code AI powered automation tool.

Obsei is a low code AI powered automation tool. It can be used in various business flows like social listening, AI based alerting, brand image analysis, comparative study and more .

Obsei 782 Dec 31, 2022
The source code and slide for my talk about the subject: unittesing in python

PyTest Talk This talk give you some ideals about the purpose of unittest? how to write good unittest? how to use pytest framework? and show you the ba

nguyenlm 3 Jan 18, 2022
A pytest plugin, that enables you to test your code that relies on a running PostgreSQL Database

This is a pytest plugin, that enables you to test your code that relies on a running PostgreSQL Database. It allows you to specify fixtures for PostgreSQL process and client.

Clearcode 252 Dec 21, 2022
A pytest plugin that enables you to test your code that relies on a running Elasticsearch search engine

pytest-elasticsearch What is this? This is a pytest plugin that enables you to test your code that relies on a running Elasticsearch search engine. It

Clearcode 65 Nov 10, 2022
This is a pytest plugin, that enables you to test your code that relies on a running MongoDB database

This is a pytest plugin, that enables you to test your code that relies on a running MongoDB database. It allows you to specify fixtures for MongoDB process and client.

Clearcode 19 Oct 21, 2022
Pytest-typechecker - Pytest plugin to test how type checkers respond to code

pytest-typechecker this is a plugin for pytest that allows you to create tests t

vivax 2 Aug 20, 2022
It's a simple script to generate a mush on code forces, the script will accept the public problem urls only or polygon problems.

Codeforces-Sheet-Generator It's a simple script to generate a mushup on code forces, the script will accept the public problem urls only or polygon pr

Ahmed Hossam 10 Aug 2, 2022
This is the official source code for SLATE. We provide the code for the model, the training code, and a dataset loader for the 3D Shapes dataset. This code is implemented in Pytorch.

SLATE This is the official source code for SLATE. We provide the code for the model, the training code and a dataset loader for the 3D Shapes dataset.

Gautam Singh 66 Dec 26, 2022
IDE allow you to refactor code, Baron allows you to write refactoring code.

Introduction Baron is a Full Syntax Tree (FST) library for Python. By opposition to an AST which drops some syntax information in the process of its c

Python Code Quality Authority 278 Dec 29, 2022
Various code metrics for Python code

Radon Radon is a Python tool that computes various metrics from the source code. Radon can compute: McCabe's complexity, i.e. cyclomatic complexity ra

Michele Lacchia 1.4k Jan 7, 2023
Code examples for my Write Better Python Code series on YouTube.

Write Better Python Code This repository contains the code examples used in my Write Better Python Code series published on YouTube: https:/

null 858 Dec 29, 2022
Django project starter on steroids: quickly create a Django app AND generate source code for data models + REST/GraphQL APIs (the generated code is auto-linted and has 100% test coverage).

Create Django App ?? We're a Django project starter on steroids! One-line command to create a Django app with all the dependencies auto-installed AND

imagine.ai 68 Oct 19, 2022
TensorFlow code for the neural network presented in the paper: "Structural Language Models of Code" (ICML'2020)

SLM: Structural Language Models of Code This is an official implementation of the model described in: "Structural Language Models of Code" [PDF] To ap

null 73 Nov 6, 2022
Code to use Augmented Shapiro Wilks Stopping, as well as code for the paper "Statistically Signifigant Stopping of Neural Network Training"

This codebase is being actively maintained, please create and issue if you have issues using it Basics All data files are included under losses and ea

Justin Terry 32 Nov 9, 2021
Inference code for "StylePeople: A Generative Model of Fullbody Human Avatars" paper. This code is for the part of the paper describing video-based avatars.

NeuralTextures This is repository with inference code for paper "StylePeople: A Generative Model of Fullbody Human Avatars" (CVPR21). This code is for

Visual Understanding Lab @ Samsung AI Center Moscow 18 Oct 6, 2022
A code generator from ONNX to PyTorch code

onnx-pytorch Generating pytorch code from ONNX. Currently support onnx==1.9.0 and torch==1.8.1. Installation From PyPI pip install onnx-pytorch From

Wenhao Hu 94 Jan 6, 2023
SCodeScanner stands for Source Code scanner where the user can scans the source code for finding the Critical Vulnerabilities.

The SCodeScanner stands for Source Code Scanner, where you can scan your source code files like PHP and get identify the vulnerabilities inside it. The tool can use by Pentester, Developer to quickly identify the weakness.

null 136 Dec 13, 2022