Inquiry about the DynGEM

[ghost]

Hi paulpjoby,
the codes of DynGEM == "DynGEM: Deep Embedding Method for Dynamic Graphs" in Arxiv?
https://arxiv.org/abs/1805.11273

Thanks.

[HxWithMushroom]

Hi paulpjoby，
i have a question about this code.
i find that the last two sentences in the file "dyngem.py" was commented out, that is "encoder = get_encoder(f_model, input_layer_name)"
when i run into the function "get_encoder()", and it goes to the sentence"encoder = Model(encoder.input, encoder.get_layer('embedding-layer').output)", it will throw an exception, says"Layer embedding-layer has multiple inbound nodes, hence the notion of "layer output" is ill-defined. Use get_output_at(node_index) instead."
I would like to ask what is the cause of this problem and how to solve it. Thanks

[paulpjoby]

Hi paulpjoby,
the codes of DynGEM == "DynGEM: Deep Embedding Method for Dynamic Graphs" in Arxiv?
https://arxiv.org/abs/1805.11273

Thanks.

Sorry ! I think you meant the expansion of DynGEM right ? I didn't get you exactly please let me know if I got you incorrectly

Thanks.

[paulpjoby]

Hi paulpjoby，
i have a question about this code.
i find that the last two sentences in the file "dyngem.py" was commented out, that is "encoder = get_encoder(f_model, input_layer_name)"
when i run into the function "get_encoder()", and it goes to the sentence"encoder = Model(encoder.input, encoder.get_layer('embedding-layer').output)", it will throw an exception, says"Layer embedding-layer has multiple inbound nodes, hence the notion of "layer output" is ill-defined. Use get_output_at(node_index) instead."
I would like to ask what is the cause of this problem and how to solve it. Thanks

Hi @HxWithMushroom
I was away from keyboard for few days. I will look into the issue and get back to you as soon as possible.

[hxtruong6]

Hi paulpjoby,
I see your loss function is not exactly following DynGEM papper. Do you use other loss function?
And you also not use mini batch for training?
Thanks