A model of targeted advertising with customer recognition

A two-period duopoly model is developed to examine the competitive effects of targeted advertising with customer recognition （TACR）. In the model, two competing firms sell goods to end consumers in the first period, during which customer recognition is obtained. In the second period, advertising can be targeted toward different consumer types. Advertising is assumed to be persuasive in the way that consumer valuation is increased. Equilibrium decisions and profits in each period are derived, showing that the firm who loses the current competition will win in the future. As a result, forward-looking firms price less aggressively so that their long-term profits can be enhanced with the help of TACR. Particularly, TACR improves profits through three important effects： valuation increasing, customer poaching, and anti-competition. Finally, this paper investigates the welfare issues, showing that firms enhance profits at the expense of consumer surplus. It is, therefore, suggested that public sectors take a step to protect consumers with the rapid development of targeting technology.

Mobile-Customer Identity Recognition

By utilizing artificial intelligence and pattern rec ognition techniques, we propose an integrated mobile-customer identity recognition approach in this paper, based on cus tomer＇s behavior characteristics extracted from the customer information database. To verify the effectiveness of this approach, a test has been run on the dataset consisting of 1 000 customers in 3 consecutive months.The result is compared with the real dataset in the fourth month consisting of 162 customers, which has been set as the customers for recognition. The high correct rate of the test （96.30%）, together with 1. 87% of the judge-by-mistake rate and 7.82% of the leaving-out rate, demonstrates the effectiveness of this approach.

Audiovisual speech recognition based on a deep convolutional neural network

Audiovisual speech recognition is an emerging research topic.Lipreading is the recognition of what someone is saying using visual information,primarily lip movements.In this study,we created a custom dataset for Indian English linguistics and categorized it into three main categories:(1)audio recognition,(2)visual feature extraction,and(3)combined audio and visual recognition.Audio features were extracted using the mel-frequency cepstral coefficient,and classification was performed using a one-dimension convolutional neural network.Visual feature extraction uses Dlib and then classifies visual speech using a long short-term memory type of recurrent neural networks.Finally,integration was performed using a deep convolutional network.The audio speech of Indian English was successfully recognized with accuracies of 93.67%and 91.53%,respectively,using testing data from 200 epochs.The training accuracy for visual speech recognition using the Indian English dataset was 77.48%and the test accuracy was 76.19%using 60 epochs.After integration,the accuracies of audiovisual speech recognition using the Indian English dataset for training and testing were 94.67%and 91.75%,respectively.