18156512. RECOMMENDATION MODEL TRAINING METHOD, RECOMMENDATION METHOD, APPARATUS, AND COMPUTER-READABLE MEDIUM simplified abstract (HUAWEI TECHNOLOGIES CO., LTD.)
RECOMMENDATION MODEL TRAINING METHOD, RECOMMENDATION METHOD, APPARATUS, AND COMPUTER-READABLE MEDIUM
Organization Name
Inventor(s)
Xiaolian Zhang of Shenzhen (CN)
RECOMMENDATION MODEL TRAINING METHOD, RECOMMENDATION METHOD, APPARATUS, AND COMPUTER-READABLE MEDIUM - A simplified explanation of the abstract
This abstract first appeared for US patent application 18156512 titled 'RECOMMENDATION MODEL TRAINING METHOD, RECOMMENDATION METHOD, APPARATUS, AND COMPUTER-READABLE MEDIUM
Simplified Explanation
The patent application describes a training method for recommendation models. Here is a simplified explanation of the abstract:
- The method involves obtaining a first recommendation model by training it on a set of training samples.
- An impact function value is determined for each training sample, indicating its influence on the verification loss of a set of second training samples.
- Based on the impact function values, weights are assigned to each training sample.
- The first recommendation model is then trained again using the training samples and their corresponding weights to obtain a target recommendation model.
Potential applications of this technology:
- Personalized recommendation systems in e-commerce platforms.
- Content recommendation algorithms for streaming services.
- Product recommendation engines for online marketplaces.
Problems solved by this technology:
- Improving the accuracy and effectiveness of recommendation models.
- Addressing the challenge of selecting the most influential training samples for model training.
- Enhancing the performance of recommendation systems in various domains.
Benefits of this technology:
- Enhanced accuracy in recommending relevant items or content to users.
- Improved efficiency in training recommendation models.
- Increased personalization and user satisfaction in recommendation systems.
Original Abstract Submitted
A training method includes: obtaining a first recommendation model, where a model parameter of the first recommendation model is obtained through training based on n first training samples; determining an impact function value of each first training sample with respect to a verification loss of m second training samples in the first recommendation model; determining, based on the impact function value of each first training sample with respect to the verification loss, a weight corresponding to each first training sample; and training the first recommendation model based on the n first training samples and the weights corresponding to the n first training samples, to obtain a target recommendation model.
