18141141. INCORPORATING STRUCTURED KNOWLEDGE IN NEURAL NETWORKS simplified abstract (Microsoft Technology Licensing, LLC)
Contents
- 1 INCORPORATING STRUCTURED KNOWLEDGE IN NEURAL NETWORKS
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 INCORPORATING STRUCTURED KNOWLEDGE IN NEURAL NETWORKS - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Key Features and Innovation
- 1.6 Potential Applications
- 1.7 Problems Solved
- 1.8 Benefits
- 1.9 Commercial Applications
- 1.10 Questions about Structured Knowledge Modeling and Neural Networks
- 1.11 Original Abstract Submitted
INCORPORATING STRUCTURED KNOWLEDGE IN NEURAL NETWORKS
Organization Name
Microsoft Technology Licensing, LLC
Inventor(s)
Bhaskar Mitra of Montreal (CA)
Yordan Kirilov Zaykov of Cambridge (GB)
John Michael Winn of Redmond (GB)
James John Hensman of Redmond (GB)
INCORPORATING STRUCTURED KNOWLEDGE IN NEURAL NETWORKS - A simplified explanation of the abstract
This abstract first appeared for US patent application 18141141 titled 'INCORPORATING STRUCTURED KNOWLEDGE IN NEURAL NETWORKS
Simplified Explanation
The patent application describes a method for structured knowledge modeling and incorporating learned knowledge in neural networks. Knowledge is stored in a knowledge base in a structured and human-interpretable manner. Another neural network can read from and write to this knowledge model using structured queries.
- Knowledge is encoded in a structured knowledge base for human interpretability and verifiability.
- Neural networks can interact with the knowledge base through structured queries.
- The knowledge model uses property name-value structures for interpretability.
- The system allows for gradient-based training and updates, enabling knowledge inference from training data.
Key Features and Innovation
- Structured knowledge modeling for human interpretability.
- Neural network interaction with knowledge base through structured queries.
- Property name-value structure for interpretability.
- Gradient-based training for knowledge inference.
Potential Applications
This technology can be applied in various fields such as:
- Natural language processing
- Information retrieval systems
- Knowledge management systems
Problems Solved
- Lack of human-interpretable knowledge representation in neural networks.
- Difficulty in incorporating learned knowledge into neural network models.
Benefits
- Improved interpretability of knowledge in neural networks.
- Enhanced ability to incorporate learned knowledge into models.
- Efficient knowledge inference from training data.
Commercial Applications
Potential commercial applications include:
- AI-powered chatbots
- Recommendation systems
- Data analysis tools
Questions about Structured Knowledge Modeling and Neural Networks
How does structured knowledge modeling improve interpretability in neural networks?
Structured knowledge modeling organizes knowledge in a human-interpretable way, making it easier for neural networks to understand and utilize the information effectively.
What are the benefits of incorporating learned knowledge into neural networks?
Incorporating learned knowledge enhances the performance and adaptability of neural networks, allowing them to make more informed decisions based on past experiences.
Original Abstract Submitted
An approach to structured knowledge modeling and the incorporation of learned knowledge in neural networks is disclosed. Knowledge is encoded in a knowledge base (KB) in a manner that is explicit and structured, such that it is human-interpretable, verifiable, and editable. Another neural network is able to read from and/or write to the knowledge model based on structured queries. The knowledge model has an interpretable property name-value structure, represented using property name embedding vectors and property value embedding vectors, such that an interpretable, structured query on the knowledge base may be formulated by a neural model in terms of tensor operations. The knowledge base admits gradient-based training or updates (of the knowledge base itself and/or a neural network(s) supported by the knowledge base), allowing knowledge or knowledge representations to be inferred from a training set using machine learning training methods.
