One of the areas in which businesses use artificial intelligence techniques is the analysis and prediction of customer behavior. It is important for a business to predict the future behavior of its customers. In this paper, a customer behavior model using wild horse optimization algorithm is proposed. In the first step, K-Means algorithm is used to classify based on the features extracted from the time series, and then in the second step, wild horse optimization algorithm is used to estimate customer behavior. Three dataset including, the grocery store dataset, the household appliances dataset, and the supermarket dataset are used in the simulation. The best clusters count for the grocery store dataset, the household appliances dataset, and the supermarket dataset are obtained 5, 4, and 4, respectively. The simulation results indicate that this proposed method is obtained the lowest prediction error in three simulated datasets and is superior to other counterparts. پرونده مقاله

In this paper, an improved decision tree classification method based on wild horse optimization algorithm is proposed and then the application in customer behavior analysis is evaluated. Customer behavior is modeled in the form of time series. The proposed method includes two general steps. First, the customers are classified into clusters based on the features extracted from the time series, and then the customers’ behavior is estimated based on an efficient predictive algorithm in the second step. In this paper, an improved decision tree classification based on wild horse optimization algorithm is used to predict customer behavior. The proposed method is implemented in the MATLAB software environment and its efficiency is evaluated in the Symmetric Mean Absolute Percentage Error (SMAPE) index. The experimental results show that variance, spikiness, lumpiness and entropy have a high impact intensity among the extracted features. The overall evaluation indicate that this proposed method obtains the lowest prediction error in compared to other evaluated methods. پرونده مقاله

Scene change detection plays an important role in a number of video applications, including video indexing, searching, browsing, semantic features extraction, and, in general, pre-processing and post-processing operations. Several scene change detection methods have been proposed in different coding standards. Most of them use fixed thresholds for the similarity metrics to determine if there was a change or not. These thresholds are obtained empirically or they must be calculated before the scene change detection after the whole sequence is obtained. Efficiency of scene change detectors decreases considerably for videos with high scene complexity and variation. In this paper, we propose a novel scene change detection algorithm in the HEVC compressed domain. In the proposed method, we have developed an efficient method based on the analysis of the Transform Units distribution in HEVC standard. In order to enhance the accuracy of detecting the scene changes, we have also defined an automated, dynamic threshold model which can efficiently trace scene changes. The experimental results on UHD videos indicate a higher performance with significantly improved accuracy combined with minimum complexity. پرونده مقاله

Intra coding in High efficiency video coding (HEVC) can significantly improve the compression efficiency using 35 intra-prediction modes for 2N×2N (N is an integer number ranging from six to two) luma blocks. To find the luma block with the minimum rate-distortion, it must perform 11932 different rate-distortion cost calculations. Although this approach improves coding efficiency compared to the previous standards such as H.264/AVC, but computational complexity is increased significantly. In this paper, an intra-prediction technique has been described to improve the performance of the HEVC standard by minimizing its computational complexity. The proposed algorithm called prediction unit size decision (PUSD) was introduced to decrease evaluation of block sizes. The simulation results show that the time complexity is decreased by ~36% while the bit-rate is increased by 1.1 kbps, and PSNR is decreased by 0.6 db. Accordingly, the proposed algorithms have negligible effect on the video quality with great saving in the time complexity. پرونده مقاله

three dimensional- high efficiency video coding (3D-HEVC) is the expanded version of the latest video compression standard, namely high efficiency video coding (HEVC), which is used to compress 3D videos. 3D videos include texture video and depth map. Since the statistical characteristics of depth maps are different from those of texture videos, new tools have been added to the HEVC standard for intra coding. Thirty-five intra prediction modes with the recursive block partitioning structure in HEVC and the depth modeling modes improve the intra coding efficiency while increase the computational complexity. This Standard achieves the highest possible coding efficiency compared with predecessor standards, while increases the computational complexity that makes the 3D-HEVC cannot be suitable for real-time applications. In this paper, a fast intra prediction mode decision method is proposed to code depth map for 3D videos. Since the texture of video and the corresponding depth map represent the same scene, there is a high correlation between the prediction mode of texture video and its depth map. Thus, we can skip some depth map intra prediction modes rarely used in the related texture coding unit. The simulation results show that the proposed method reduces computational complexity by 23.66% compared to 3D-HEVC standard with an increase of 0.09% bit rate پرونده مقاله