A Hybrid Model for Commercial Brand Marketing Prediction Based on Multiple Features with Image Processing

Recently, deep learning has been employed in automatic feature extraction and made remarkable achievements the fields of computer vision, speech recognition, natural language processing, artificial intelligence. Compared with traditional shallow model, can automatically extract more complex features from simple features, which reduces intervention engineering to a certain extent. With development Internet e-commerce, picture advertising, as an important form display characteristics high visibility, strong readability, easy-to-obtain user recognition. An increasing number companies are paying attention what kind advertising pictures attract clicks. Based on technology, this paper studies prediction model click-through rate (CTR) for proposes end-to-end CTR depth integrates directly predict probability advertisement image being clicked by users. This deep-seated nonlinear through multilayer network structure carries out several groups experiments private data set commercial platform. The results show that proposed effectively improve accuracy compared other benchmark models whether is or not given information information. By establishing reasonable it help platform estimate future revenue so make cooperative decisions advertisers. For advertisers, necessary evaluate price predicting bidding their own advertisements.