RECORD
Incorporating Biological Knowledge with Factor Graph Neural Network for Interpretable Deep Learning
Genomics · Submitted 3 Jun 2019
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim We applied our model to two cancer genomic datasets to predict target clinical variables and achieved better results than other traditional machine learning and deep learning models.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
While deep learning has achieved great success in many fields, one common criticism about deep learning is its lack of interpretability. In most cases, the hidden units in a deep neural network do not have a clear semantic meaning or correspond to any physical entities. However, model interpretability and explainability are crucial in many biomedical applications. To address this challenge, we developed the Factor Graph Neural Network model that is interpretable and predictable by combining probabilistic graphical models with deep learning. We directly encode biological knowledge such as Gene Ontology as a factor graph into the model architecture, making the model transparent and interpretable. Furthermore, we devised an attention mechanism that can capture multi-scale hierarchical interactions among biological entities such as genes and Gene Ontology terms. With parameter sharing mechanism, the unrolled Factor Graph Neural Network model can be trained with stochastic depth and generalize well. We applied our model to two cancer genomic datasets to predict target clinical variables and achieved better results than other traditional machine learning and deep learning models. Our model can also be used for gene set enrichment analysis and selecting Gene Ontology terms that are important to target clinical variables.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | While deep learning has achieved great success in many fields, one common criticism about deep learning is its lack of interpretability. | background 0.99 | 0.04 | 0.02 | asserted 0.99 |
| s1 | In most cases, the hidden units in a deep neural network do not have a clear semantic meaning or correspond to any physical entities. | background 0.97 | 0.06 | 0.02 | asserted 0.96 |
| s2 | However, model interpretability and explainability are crucial in many biomedical applications. | background 0.93 | 0.04 | 0.03 | asserted 1.00 |
| s3 | To address this challenge, we developed the Factor Graph Neural Network model that is interpretable and predictable by combining probabilistic graphical models with deep learning. | contribution 0.69 | 0.96 | 0.18 | asserted 1.00 |
| s4 | We directly encode biological knowledge such as Gene Ontology as a factor graph into the model architecture, making the model transparent and interpretable. | method 0.88 | 0.96 | 0.11 | asserted 1.00 |
| s5 | Furthermore, we devised an attention mechanism that can capture multi-scale hierarchical interactions among biological entities such as genes and Gene Ontology terms. | method 0.72 | 0.96 | 0.05 | asserted 0.96 |
| s6 | With parameter sharing mechanism, the unrolled Factor Graph Neural Network model can be trained with stochastic depth and generalize well. | method 0.97 | 0.85 | 0.05 | asserted 0.97 |
| s7 | We applied our model to two cancer genomic datasets to predict target clinical variables and achieved better results than other traditional machine learning and deep learning models. | result 0.99 | 0.97 | 0.48 | asserted 1.00 |
| s8 | Our model can also be used for gene set enrichment analysis and selecting Gene Ontology terms that are important to target clinical variables. | contribution 0.70 | 0.90 | 0.16 | asserted 0.67 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
While deep learning has achieved great success in many fields, one common criticism about deep learning is its lack of interpretability. In most cases, the hidden units in a deep neural network do not have a clear semantic meaning or correspond to any physical entities. However, model interpretability and explainability are crucial in many biomedical applications. To address this challenge, we developed the Factor Graph Neural Network model that is interpretable and predictable by combining probabilistic graphical models with deep learning. We directly encode biological knowledge such as Gene Ontology as a factor graph into the model architecture, making the model transparent and interpretable. Furthermore, we devised an attention mechanism that can capture multi-scale hierarchical interactions among biological entities such as genes and Gene Ontology terms. With parameter sharing mechanism, the unrolled Factor Graph Neural Network model can be trained with stochastic depth and generalize well. We applied our model to two cancer genomic datasets to predict target clinical variables and achieved better results than other traditional machine learning and deep learning models. Our model can also be used for gene set enrichment analysis and selecting Gene Ontology terms that are important to target clinical variables.
Human review
References
How the literature uses it
5 arXiv papers cite this in 5 sentences
What it cites · 5 sentences without a confident label
Builds on 1 paper
-
It generalizes the Graph Convolutional Network (GCN) [ref] and can capture hierarchical multi-scale interactions with attention mechanisms.
Semi-Supervised Classification with Graph Convolutional Networks · Related work · 0.91
Compares with 1 paper
-
We compared our methods with traditional machine learning methods Random Forest, Decision Tree and Multilayer Perceptron (MLP), as well as two recent deep learning methods including Graph Convolutional Neural Network (GCN) [ref] , and Prototypical Network [ref] .
Semi-Supervised Classification with Graph Convolutional Networks · Data preprocessing · 0.98
Cites as background 2 papers
-
As graphs are commonly used in many applications, novel deep learning models on graphs have been developed, including Graph Neural Network [ref] , Graph Convolutional Neural Network [ref] , and so on.
Semi-Supervised Classification with Graph Convolutional Networks · Related work · 0.93
-
Graph Attention Model (GAM) [ref] also applied the attention mechanism to learn graph embeddings.
Graph Attention Networks · Related work · 0.93
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.