17690078. ELECTRONIC APPARATUS AND METHOD WITH UNCERTAINTY ESTIMATION IMPLEMENTING DATA LABELS simplified abstract (Samsung Electronics Co., Ltd.)
ELECTRONIC APPARATUS AND METHOD WITH UNCERTAINTY ESTIMATION IMPLEMENTING DATA LABELS
Organization Name
Inventor(s)
Junhwi Choi of Seongnam-si (KR)
ELECTRONIC APPARATUS AND METHOD WITH UNCERTAINTY ESTIMATION IMPLEMENTING DATA LABELS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17690078 titled 'ELECTRONIC APPARATUS AND METHOD WITH UNCERTAINTY ESTIMATION IMPLEMENTING DATA LABELS
Simplified Explanation
The patent application describes a training method that involves using prediction models and uncertainty models to improve the accuracy of predictions. Here are the key points:
- The method starts by generating prediction data using an auxiliary prediction model, based on input data.
- A primary uncertainty model is then used to estimate the uncertainty associated with the input data and its corresponding label.
- A loss is determined based on the generated prediction data, the label, and the estimated uncertainty.
- The primary uncertainty model is trained using this loss.
- Next, prediction data is generated using a primary prediction model, based on different input data.
- The trained primary uncertainty model is used to estimate the uncertainty associated with this new input data and its corresponding label.
- Another loss is determined based on the generated prediction data, the label, and the estimated uncertainty.
- The primary prediction model is then trained using this second loss.
Potential applications of this technology:
- Improving the accuracy of prediction models in various fields such as finance, healthcare, and weather forecasting.
- Enhancing the performance of machine learning algorithms in tasks like image recognition, natural language processing, and recommendation systems.
Problems solved by this technology:
- Addressing the issue of uncertainty in prediction models, which can lead to inaccurate or unreliable results.
- Providing a systematic approach to training models that takes into account uncertainty estimation, leading to more robust and reliable predictions.
Benefits of this technology:
- Increased accuracy and reliability of prediction models.
- Improved decision-making based on more accurate predictions.
- Better understanding and management of uncertainty in machine learning algorithms.
Original Abstract Submitted
A training method is provided. The training method may include generating first prediction data based on first data by inputting the first data to a trained auxiliary prediction model, estimating a first uncertainty by inputting the first data and a first label to a primary uncertainty model, determining a first uncertainty loss based on the generated first prediction data, the first label, and the estimated first uncertainty, training the primary uncertainty model based on the determined first uncertainty loss, generating second prediction data based on second data by inputting the second data to a primary prediction model, estimating a second uncertainty by inputting the second data and a second label to the trained primary uncertainty model, determining a second uncertainty loss based on the generated second prediction data, the second label, and the estimated second uncertainty, and training the primary prediction model based on the determined second uncertainty loss.
