18443711. RAPID ESTIMATION OF THE UNCERTAINTY OF THE OUTPUT OF A NEURAL TASK NETWORK simplified abstract (Robert Bosch GmbH)
Contents
RAPID ESTIMATION OF THE UNCERTAINTY OF THE OUTPUT OF A NEURAL TASK NETWORK
Organization Name
Inventor(s)
Lydia Gauerhof of Sindelfingen (DE)
Christoph Schorn of Benningen Am Neckar (DE)
RAPID ESTIMATION OF THE UNCERTAINTY OF THE OUTPUT OF A NEURAL TASK NETWORK - A simplified explanation of the abstract
This abstract first appeared for US patent application 18443711 titled 'RAPID ESTIMATION OF THE UNCERTAINTY OF THE OUTPUT OF A NEURAL TASK NETWORK
- Simplified Explanation:**
This patent application describes a method for training a measurement network to determine the uncertainty of an already trained task network. Training data is fed to multiple versions of the task network or multiple times to a probabilistic task network to generate various outputs. The measurement network then predicts the distribution of these outputs and evaluates the consistency with a predefined cost function.
- Key Features and Innovation:**
- Training a measurement network to assess uncertainty in a trained task network.
- Utilizing multiple versions of a deterministic task network or multiple runs of a probabilistic task network to generate diverse outputs.
- Predicting the distribution of outputs using the measurement network.
- Evaluating the consistency of the prediction with a predefined cost function.
- Optimizing network parameters to enhance the measurement network's behavior.
- Potential Applications:**
This technology can be applied in various fields such as finance, healthcare, weather forecasting, and quality control to assess the uncertainty of predictions made by task networks.
- Problems Solved:**
This technology addresses the need for accurate uncertainty estimation in trained task networks, which is crucial for making informed decisions based on the predictions generated.
- Benefits:**
- Improved accuracy in uncertainty estimation.
- Enhanced decision-making based on more reliable predictions.
- Increased trust in the outputs of task networks.
- Commercial Applications:**
Potential commercial applications include financial risk assessment, medical diagnosis support systems, predictive maintenance in manufacturing, and optimizing supply chain management.
- Questions about the Technology:**
1. How does this method improve the reliability of predictions made by task networks? 2. What are the key advantages of using a measurement network to assess uncertainty in trained task networks?
Original Abstract Submitted
A method for training a measurement network which ascertains the uncertainty of an already trained task network. In the method: each training record of measurement data from a training data set is fed to a plurality of modifications of a deterministic task network, or fed multiple times to a probabilistic task network, and thus mapped onto a plurality of outputs; each training record is fed to the measurement network and mapped onto a prediction of the distribution of the plurality of outputs, wherein the processing chain of the measurement network includes a part of the processing chain of the task network; a predefined cost function evaluates the extent to which the prediction of the distribution is consistent with the outputs; and network parameters which characterize the behavior of that part of the measurement network that does not belong to the processing chain of the task network are optimized.
