Feature extraction and model acquisition method and device, electronic equipment, medium and product
By constructing a graph G(V, A, F) and training an encoding-decoding model, the problems of feature explosion and non-universal encoding in the processing of geolocation information in ride-hailing and food delivery platforms are solved, and efficient feature extraction applicable to multiple models is achieved.
CN116662679BActive Publication Date: 2026-07-24ALIBABA (CHINA) CO LTD
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ALIBABA (CHINA) CO LTD
- Filing Date
- 2022-02-15
- Publication Date
- 2026-07-24
Smart Images
Figure CN116662679B_ABST
Abstract
Embodiments of the present disclosure disclose a feature extraction and model acquisition method and device, electronic equipment, medium and product. The method comprises: constructing a graph G(V, A, F) based on the connection relationship between each grid region and the historical behavior feature vector of each grid region; the nodes V in the graph represent each grid region; the edges in the graph represent the connection relationship between each grid region, denoted as matrix A; the feature vector F of each node in the graph is the historical behavior feature vector of each grid region; and training a coding-decoding model based on the graph G(V, A, F) to obtain a feature extraction model, wherein the feature extraction model is used to extract the behavior geographical location feature vector of each node in the graph. The behavior geographical location feature vector extracted by the technical solution can be widely applied to various deep models or non-deep models that require input of behavior geographical location features.
Need to check novelty before this filing date? Find Prior Art