Questions tagged [graph-neural-network]
The graph-neural-network tag has no summary.
54 questions
5 votes
0 answers
51 views
Why are "per-sample graphs" rarely studied in GNN research?
I've been diving into Graph Neural Networks lately, and I've noticed that most papers seem to focus on scenarios where all samples share a single, large graph — like citation networks or social graphs....
- 51
1 vote
0 answers
38 views
How to correctly use a transformer model for a generating dependencies project
I'm currectly trying to train a model in order to predict dependencies between text, here it's industrials tasks, based on historical data. The goal is to learn that "Task A precedes Task B for ...
- 361
5 votes
1 answer
94 views
Choosing NN architecture for route correction
Can you help me choose a NN architecture that solves this kind of problem: A route, meaning a sequence of coordinates (e.g. [(34.1, 52.3),(34.101, 52.301), ... ]), ...
3 votes
1 answer
181 views
Pytorch mat1 and mat2 shapes cannot be multiplied (131072x12 and 64x32)
I have a multivariate time series forecasting model that originally used dilated temporal convolution for temporal dependencies extraction that i tried to replace with multi head self-attention with ...
- 31
1 vote
0 answers
23 views
Could I use GCN get nodes structual embedding without training?
Now, I want to get nodes' structual embedding, use DeepWalk or Node2Vec couldn't get same vector even they have same structure. then I use GCN which could get same vector without training. So I want ...
- 11
0 votes
1 answer
23 views
Improving Data Sets by Adding New Nodes
I have an idea for adding new nodes into a directed graph to achieve a numerically stable graph Fourier transform for signal analysis in the graph domain. To assess its efficiency, I am looking for ...
- 103
2 votes
0 answers
31 views
spectral gnn forward pass
Following the article: A Practical Guide to Graph Neural Networks: https://arxiv.org/abs/2010.05234 Suppose our graph has $N$ nodes and Laplacian $L = UDU^T$. Let $\Theta$ denote a filter. Let $f_{k}$ ...
- 121
1 vote
1 answer
67 views
How can a citation dataset (like Cora) have strongly connected components?
This website https://snap.stanford.edu/data/cit-HepPh.html shows that the High-energy physics citation network has strongly connected components and it's driving me crazy. A SCC would mean that you ...
- 111
1 vote
1 answer
66 views
Asynchronous Training of Deep Learning Models
I am thinking of how would it be if I can create asynchronous forward function in sub-class of nn.Module . When I came across architecture in attached image, I felt that it would be faster if we could ...
- 13
1 vote
1 answer
367 views
What're the standard ways of padding data for GNNs?
I am working on materials property prediction using GNNs with torch_geometric. Each data in my dataset has different number of feature vectors x, edge_index vectors ...
- 11
0 votes
0 answers
300 views
Convert specific domain knowledge text to a knowledge graph
As part of this semester assignment , I'm working on a project that aims to to represent the knowledge in "PMBOK 6th edition, section 11: Project Risk Management (page 395 -> 458)" and the knowledge ...
0 votes
1 answer
2k views
Implementation of Graph Neural Network for Image Classification
I'm currently working on a project where I want to utilize Graph Neural Networks (GNNs) for image classification tasks. However, I'm facing difficulties in understanding how to implement GNNs ...
- 170
2 votes
1 answer
405 views
Trouble Training GNN for Binary Node Classification Task
I am using a GNN to solve a problem in which I have a query target and an undirected graph. My goal is to emit a subset of nodes in the graph (via a node-wise binary prediction) whose features sum to ...
- 21
0 votes
1 answer
56 views
What are effective ways to merge different parts of a document to form a single document?
I have a dataset where I have 3 different columns (Title, Keyword and Abstract) representing a document. I have to build a text classification model using TextGCN, where documents and words will be ...
0 votes
1 answer
275 views
How to correctly implement the NNConv in Pytorch?
I tried to make a GNN class which can make use of my node features along with my edge features of the graph.I have implemented NNConv in order to use the edge features, but I am not able to understand ...
- 103