International business machines corporation (20240104366). MULTIPLEXED GRAPH NEURAL NETWORKS FOR MULTIMODAL FUSION simplified abstract

MULTIPLEXED GRAPH NEURAL NETWORKS FOR MULTIMODAL FUSION

Organization Name

international business machines corporation

Inventor(s)

Niharika Dsouza of San Jose CA (US)

Tanveer Syeda-mahmood of Cupertino CA (US)

Andrea Giovannini of Zurich (CH)

Antonio Foncubierta Rodriguez of Zurich (CH)

MULTIPLEXED GRAPH NEURAL NETWORKS FOR MULTIMODAL FUSION - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240104366 titled 'MULTIPLEXED GRAPH NEURAL NETWORKS FOR MULTIMODAL FUSION

Simplified Explanation

The patent application describes a method for transforming data samples into a multiplexed graph using a graph neural network layer with different connectivity patterns between nodes.

• The method involves transforming data samples into a multiplexed graph by creating planes with nodes and edges associated with different relation types.
• Message passing walks are performed within and across the planes using a graph neural network layer with units that output aggregations of sub-units.
• The sub-units include a typed GNN layer that allows for different connectivity patterns between intra-planar and inter-planar nodes.
• Task-specific supervision is used to train the weights of the GNN for the machine learning task.

Potential Applications

This technology can be applied in various fields such as:

• Social network analysis
• Recommendation systems
• Bioinformatics

Problems Solved

• Efficiently transforming data samples into a multiplexed graph
• Handling different relation types in the graph
• Improving message passing walks within and across planes

Benefits

• Enhanced graph representation of data
• Improved performance in machine learning tasks
• Flexibility in handling different connectivity patterns

Potential Commercial Applications

Optimized for SEO: "Graph Neural Network Technology for Data Analysis and Machine Learning"

• Data analytics companies
• Machine learning software developers
• Research institutions

Possible Prior Art

There may be prior art related to graph neural networks, multiplexed graphs, and message passing algorithms in the field of machine learning and data analysis.

How does this method compare to existing graph neural network approaches in terms of performance and scalability?

This article does not provide a direct comparison with existing graph neural network approaches in terms of performance and scalability. Further research or experimentation may be needed to address this question.

What are the potential limitations or challenges in implementing this method in real-world applications?

The article does not discuss potential limitations or challenges in implementing this method in real-world applications. Additional studies or practical implementations may be required to identify and address these issues.

Original Abstract Submitted

a computer implemented method includes transforming a set of received samples from a set of data into a multiplexed graph, by creating a plurality of planes, each plane having the set of nodes and the set of edges. each set of edges is associated with a given relation type from the set of relation types. message passing walks are alternated within and across the plurality of planes of the multiplexed graph using a graph neural network (gnn) layer. the gnn layer has a plurality of units where each unit outputs an aggregation of two parallel sub-units. sub-units include a typed gnn layer that allows different permutations of connectivity patterns between intra-planar and inter-planar nodes. a task-specific supervision is used to train a set of weights of the gnn for the machine learning task.