A video user grouping method and device based on improved kmeans algorithm
A user grouping and video technology, which is applied in computing, computer components, character and pattern recognition, etc., can solve the problems of long grouping response time, high algorithm complexity, and large computing overhead, so as to shorten the grouping response time and improve computing power. Efficiency and the effect of reducing computational overhead
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0038] like figure 1 Shown is a kind of video user grouping method based on improved Kmeans algorithm provided by the present invention, comprising:
[0039] S1. Store video user data in partitions;
[0040] S2, perform feature extraction on the video user data of each partition, input the extracted feature into the improved local Kmeans algorithm, and obtain the grouping result of each partition;
[0041] S3. Perform feature extraction for each group of each partition, and input the extracted features of all groups into the improved local Kmeans algorithm again to obtain the grouping results of all users.
[0042] The improved Partial-Kmeans algorithm proposed by the present invention and the partitioned two-layer user grouping scheme are described in detail below.
[0043] Improved Kmeans Algorithm (Partial-Kmeans)
[0044]Aiming at the problem that the calculation complexity of the Kmeans algorithm is relatively large when the amount of data is relatively large, the pres...
Embodiment 2
[0058] like figure 2 Shown, be a kind of video user grouping method based on improved Kmeans algorithm in the embodiment of the present invention, comprising:
[0059] Step 101, one or more servers form a group, and each server group uniformly reports user operation data to the same place for storage, so as to realize partition storage of user data;
[0060] Step 102, perform statistics on user data in each partition, and extract feature vectors for each user. For example, the statistics of the user's viewing records and operation behaviors within a certain period of time can be directly used as the characteristics of the user, or can be used as the characteristics of the user after certain preprocessing;
[0061] Step 103, input the users and their eigenvectors in each partition into the Partial-Kmeans algorithm for grouping, and obtain the grouping result of users in each partition.
[0062] Step 104, extract a feature vector for each user group output by each partition, ...
Embodiment 3
[0066] like image 3 As shown, the present invention provides a kind of video user grouping device based on the improved Kmeans algorithm, comprising a partition storage unit, a first-level grouping unit and a second-level grouping unit, wherein:
[0067] The partition storage unit is used to partition and store video user data;
[0068] The first-level grouping unit is used to extract features from the video user data of each partition, and input the extracted features into the improved local Kmeans algorithm to obtain the grouping result of each partition;
[0069] The secondary grouping unit is used to perform feature extraction on each group of each partition, and re-input the extracted features of all groups into the improved local Kmeans algorithm to obtain the grouping results of all users.
[0070] Further, in the improved local Kmeans algorithm, the calculation formula of the center point of the kth grouping is:
[0071]
[0072] Among them, μ k is the center po...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More