18253416. Improved Processing of Sequential Data via Machine Learning Models Featuring Temporal Residual Connections simplified abstract (GOOGLE LLC)
Contents
- 1 Improved Processing of Sequential Data via Machine Learning Models Featuring Temporal Residual Connections
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 Improved Processing of Sequential Data via Machine Learning Models Featuring Temporal Residual Connections - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Original Abstract Submitted
Improved Processing of Sequential Data via Machine Learning Models Featuring Temporal Residual Connections
Organization Name
Inventor(s)
Liangzhe Yuan of Los Angeles CA (US)
Yongzhe Wang of San Francisco CA (US)
Improved Processing of Sequential Data via Machine Learning Models Featuring Temporal Residual Connections - A simplified explanation of the abstract
This abstract first appeared for US patent application 18253416 titled 'Improved Processing of Sequential Data via Machine Learning Models Featuring Temporal Residual Connections
Simplified Explanation
The patent application describes a system and method that uses a machine-learned model, such as a convolutional neural network, with temporal residual connections. These connections allow the model to transfer intermediate feature data from one instantiation of the model to another, enabling the processing of sequential inputs.
- The machine-learned model includes one or more temporal residual connections.
- Each temporal residual connection supplies intermediate feature data from the current instantiation of the model to other instantiations of the model.
- The other instantiations can process subsequent or preceding sequential inputs.
- This technology can be applied to various sequential input processing tasks.
Potential Applications
- Image recognition and classification
- Natural language processing
- Speech recognition
- Video analysis and processing
Problems Solved
- Efficient processing of sequential inputs
- Improved accuracy in sequential input analysis
- Handling of temporal dependencies in data
Benefits
- Enhanced performance in sequential input processing
- Better utilization of intermediate feature data
- Improved accuracy and efficiency in machine learning models
Original Abstract Submitted
Systems and methods can include or leverage a machine-learned model (e.g., a convolutional neural network) that includes one or more temporal residual connections. In particular, each temporal residual connection can respectively supply one or more sets of intermediate feature data generated by a current instantiation of the model from a current sequential input to one or more other instantiations of the machine-learned model applied to process one or more other sequential inputs. For example, the other instantiations of the machine-learned model can include subsequent instantiations of the machine-learned model applied to process one or more subsequent sequential inputs that follow the current sequential input in a sequence and/or preceding instantiations of the machine-learned model applied to process one or more preceding sequential inputs that precede the current sequential input in a sequence.