GNN are a special class of neural networks that are capable of working with data that is represented in graph form. These networks are heavily motivated by Convolutional Neural Networks and graph embedding. CNN’s are not able to handle graph data because the nodes in the graphs aren’t represented in any order and the fact that dependency information between two nodes is represented by edges.
Questions tagged [gnn]
56 questions
0
votes
0 answers
If graphSAGE can be converted to solve the regression problem?
I have tried to solve the node classification problems by graphSAGE and it works well on dataset Cora. Now I want to convert it to solve the regression problem, and I only change the loss function into torch.nn.MSE() without any change to other…
Echo Z
- 1
0
votes
0 answers
How do I train a TensorFlow Graph Neural Network (GNN)?
I'm attempting to write a Graph Neural Network (GNN) to solve the Time Difference of Arrival (TDoA) multilateration problem. The problem statement is as follows: given the known coordinates of 4 sensors, and the complete set of pairwise arrival time…
MomentumEigenstate
- 35
- 3
0
votes
0 answers
Memory requirement of GNN SageConv
Sorry for my simple question. My question is that in GraphSage paper, they talked about graphsage being inductive. Is Graphsage needs all of its training time samples in memory during testing? If not, how does it performs test? Since in its…
user85361
- 177
- 1
- 9
0
votes
0 answers
CF-GNNExplainer: No perturbation on 3rd GCN-layer in the given model architecture
My question refers to the official implementation of CF-GNNExplainer on Github: https://github.com/a-lucic/cf-gnnexplainer
When the perturbed model is built, the first 2 ConvLayers are the newly created perturbed layers while the 3rd layer is a…
Patrick
- 1
0
votes
0 answers
(Very Intriguing Trial) GraphSAGE: Feature Request | Feature Instruction: Exploring the Feasibility of Implementing Asymmetric Node Feature Slicing
Currently, I am working on implementing GraphSage with DGL for an industrial node classification application. Here is a brief description of my scenario:
In my application, each node in the graph has around 21 features associated with it. However,…
billz
- 11
- 1
0
votes
0 answers
GINEConv for link prediction
I have a complete graph with N nodes and (N*N-1)/2 links. Total links with positive label are N-1. I use GINEConv for link prediction. I need to predict which links are part of the graph (positive label). Since the number of positive links is much…
0
votes
0 answers
Apply Hinge Loss & Low-Rank Positives loss with Graph Neural Network for Recommendation System
I have a pytorch model I built for recommending best restaurants to user. I used graphSage and MSE as a loss between the predicted labels and the actual ones.
I want to apply the same loss mentioned on Food Discovery with Uber Eats: Using Graph…
Bushr Haddad
- 1
- 6
0
votes
0 answers
Loading a SNAP dataset into torch_geometric.data.Data object
I am trying to load a SNAP dataset (ego-facebook for now) into a torch_geometric.data.Data object. I am having issues.
Is there any standard library/method to load SNAP dataset into a torch_geometric.data.Data object?
I tried using…
H.Alzy
- 310
- 4
- 10
0
votes
0 answers
What software can extract data from image plots and output in array format?
tl;dr extracting data from image plots to data arrays
Hello all,
I am currently looking for software (either open source code or commercial software) that receives as an input an image (.jpg or .png for example) of a graph plot (for example the plot…
0
votes
0 answers
LSTM is not taking features after convolutional layer of GCN
I am trying to train Graph Convolutional Networks (GCN) and LSTM. I am facing an error when trying to pass data to LSTM after the convolutional layer of GCN. My complete pipeline is:
import torch
import torch.nn as nn
from torch_geometric.nn import…
Asif
- 763
- 8
- 18
0
votes
0 answers
GNN/Multi Agent RL in RayRLLib and PyTorch Geometric
Currently trying to add a custom_model on ray rllib, for a GNN using PyTorch. Using multi agent reinforcement learning (Specifically PPO).
The scenario is inventory management where I model each node in my supply chain as an agent. I define a…
Niki
- 1
- 1
0
votes
0 answers
How can I improve my GCN model for graph matching/entity alignment
I'm currently trying to create a very basic graph matching/ entity alignment model but the results are really not good on the train set and even worse on the test set. I also tried to add dropout to improve my results but they became even worse
So I…
Miliia
- 83
- 1
- 3
0
votes
0 answers
Questions regarding Relational Graph Convolutional Network for a Fraud Detection problem
I am currently working on a transaction dataset (https://github.com/IBM/TabFormer/tree/main/data/credit_card) and I intend to build a fraud detection engine, but with tabular data transformed into a graph. I have used this article as my main outline…
Hai Nguyen
- 13
- 1
- 2
0
votes
0 answers
AttributeError: module 'networkx' has no attribute 'from_scipy_sparse_matrix'
AttributeError Traceback (most recent call last)
in | ()
1 fig, ax = plt.subplots(figsize=(8, 8))
----> 2 ax = draw_graph(A, ax=ax, size_factor=1)
… |
SwDL
- 1
- 2
0
votes
0 answers
How to solve the OOM problem encountered when using dgl to process ultra large image data？
I am trying to generate mini batch data using a sampler with built-in support in DataLoader to support random batch training on large images,
dgl.dataloading.DataLoader( graph, valid_nids, sampler, batch_size=1024, shuffle=False, drop_last=False,…
