17850241. AUTOMATICALLY GENERATING MACHINE-LEARNING TRAINING DATA simplified abstract (Ford Global Technologies, LLC)
AUTOMATICALLY GENERATING MACHINE-LEARNING TRAINING DATA
Organization Name
Inventor(s)
Robert Parenti of Dearborn MI (US)
Adil Nizam Siddiqui of Farmington Hills MI (US)
Cynthia M. Neubecker of Westland MI (US)
AUTOMATICALLY GENERATING MACHINE-LEARNING TRAINING DATA - A simplified explanation of the abstract
This abstract first appeared for US patent application 17850241 titled 'AUTOMATICALLY GENERATING MACHINE-LEARNING TRAINING DATA
Simplified Explanation
The abstract describes a computer system that uses environmental and non-environmental data recorded by sensors on a vehicle to train a machine-learning program. The program processes environmental data by using the recorded data as training data and the annotations derived from the non-environmental data as ground truth.
- The computer system includes a processor and memory.
- Environmental data is recorded by an environmental sensor on the vehicle.
- Non-environmental data is also recorded on the vehicle independently of the environmental data.
- The system adds annotations derived from the non-environmental data to the environmental data.
- A machine-learning program is trained to process second environmental data.
- The first environmental data is used as training data for the machine-learning program.
- The annotations serve as ground truth for the first environmental data.
Potential Applications
- Autonomous vehicles: The technology can be used to train machine-learning algorithms for autonomous vehicles, improving their ability to understand and navigate the environment.
- Environmental monitoring: The system can be applied to analyze environmental data collected by sensors in various settings, such as weather monitoring or pollution detection.
- Predictive maintenance: By analyzing environmental data and correlating it with non-environmental data, the system can help predict and prevent potential vehicle failures or maintenance needs.
Problems Solved
- Lack of labeled ground truth data: The system addresses the challenge of obtaining accurate ground truth data for training machine-learning algorithms by using annotations derived from non-environmental data.
- Integration of environmental and non-environmental data: The system combines different types of data recorded on a vehicle to provide a comprehensive understanding of the environment, enabling more accurate analysis and predictions.
Benefits
- Improved machine-learning training: By using annotations derived from non-environmental data, the system enhances the training process, leading to more accurate and reliable machine-learning models.
- Enhanced environmental analysis: The integration of environmental and non-environmental data allows for a deeper understanding of the environment, enabling better decision-making and analysis.
- Predictive capabilities: The system's ability to process environmental data and predict outcomes can be leveraged in various applications, such as autonomous vehicles or predictive maintenance, leading to increased safety and efficiency.
Original Abstract Submitted
A computer includes a processor and a memory, and the memory stores instructions executable by the processor to receive first environmental data recorded by an environmental sensor on board a vehicle, receive nonenvironmental data recorded on board the vehicle independently of the first environmental data, add a plurality of annotations derived from the nonenvironmental data to the environmental data, and train a machine-learning program to process second environmental data by using the first environmental data as training data and the annotations as ground truth for the first environmental data.
