alge24 / eigenpooling Goto Github PK
View Code? Open in Web Editor NEWAn implementation of KDD paper "Graph Convolutional Networks with EigenPooling"
An implementation of KDD paper "Graph Convolutional Networks with EigenPooling"
Dear Author,
I am a follower of your paper. Thanks so much for your excellent work and I will definitely cite it if I can make some progress in this direction. Could you please kindly help me with the parameters to reproduce? I run two times of the Enzyme. One result is 65% as reported, but the other is 58.8%, it's not that stable but within reasonable range. Furthermore, could you please kindly share your parameters for other datasets? It seems that only the Enzyme dataset's parameters are provided in the .sh file.
when i run the ./run_example.sh, I get the following:
Namespace(batch_size=30, bias=True, bmname='ENZYMES', bn=True, clip=2.0, con_final=1, concat=1, datadir='data', device='cpu', dropout=0.5, feat='node-feat', feature_type='default', hidden_dim=128, input_dim=10, lr=0.001, mask=1, max_nodes=130, min_nodes=0, norm='l2', normalize=0, num_classes=6, num_epochs=900, num_gc_layers=5, num_pool_final_matrix=1, num_pool_matrix=1, num_shuffle=10, num_workers=1, out_dir='results', output_dim=128, pool_sizes='10', pred_hidden='256', shuffle=0, test_ratio=0.1, train_ratio=0.8, weight_decay=0.0, with_test=1)
bmname: ENZYMES
num_classes: 6
batch_size: 30
num_pool_matrix: 1
num_pool_final_matrix: 1
epochs: 900
learning rate: 0.001
num of gc layers: 5
output_dim: 128
hidden_dim: 128
pred_hidden: 256
dropout: 0.5
weight_decay: 0.0
shuffle: 0
Using batch normalize: True
Using feat: node-feat
Using mask: 1
Norm for eigens: l2
With test: 1
Device: cpu
Files exist, reading from stored files....
Reading file from data/data_preprocessed/ENZYMES/pool_sizes_10_nor_0/
Data loaded!
Using node features
Test ratio: 0.1
Train ratio: 0.8
Num training graphs: 480 ; Num validation graphs: 60 ; Num testing graphs: 60
Number of graphs: 600
Number of edges: 37282
Max, avg, std of graph size: 125 , 32.46 , 14.87
Traceback (most recent call last):
File "train.py", line 714, in <module>
main()
File "train.py", line 709, in main
benchmark_task_val(prog_args, pred_hidden_dims = pred_hidden_dims, feat = prog_args.feat, device=device)
File "train.py", line 434, in benchmark_task_val
prepare_data(graphs, graphs_list, args, test_graphs = None,max_nodes=args.max_nodes, seed = i)
File "train.py", line 275, in prepare_data
dataset_sampler = GraphSampler(train_graphs,train_graphs_list, args.num_pool_matrix,args.num_pool_final_matrix,normalize=False, max_num_nodes=max_nodes,
File "/home/anthony/School/Research/Projects/eigenpooling/graph_sampler.py", line 30, in __init__
self.feat_dim = G_list[0].node[0]['feat'].shape[0]
AttributeError: 'Graph' object has no attribute 'node'
Can you explain the issue? Keep in mind I am an engineer inexperienced in computer science.
Hello,
I am really interested in your excellent work, and i have some questions about the reproduction. Why the results on Gpu is sub-optimal compare to cpu which is quiet odd?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.