Research On Image Sentiment Distribution Prediction

The rapid development of mobile Internet and hardware acquisition equipment is engendering an exponential explosion of multimedia data.The way people express their opinions is also transformed from words into images.Traditional sentiment analysis methods mainly focus on predicting the most dominant sentiment category of images while neglecting the sentiment ambiguity problem restricted by various factors such as environment,subjectivity,and cultural background.To tackle this problem,visual sentiment distribution prediction has been put forward to characterize images by distributions over a set of sentiment labels instead of a single distinct label or multiple distinct label.From this perspective,this paper proposes two models of emotional distribution prediction.1.A structured low-rank inverse-covariance estimation algorithm for visual sentiment distribution prediction:The proposed model incorporates low-rank and inverse covariance regularization terms into a unified framework to learn more robust feature representation and more reasonable prediction model simultaneously.In particular,low-rank regularization term plays a pivotal role for capturing the lowrank structure embedded in data.Inverse-covariance regularization term is introduced to enforce the structured sparsity of regression coeffcients by taking the multi-output structure into account.We also develop an alternative heuristic optimization algorithm to optimize our objective function.Experiment results on three publicly available datasets,using six measurements demonstrate the superior prediction.2.A novel supervised visual sentiment distribution prediction model,termed as lowrank regularized multi-view inverse-covariance estimation.The method of Multimedia content analysis,depending on a single feature,is often difficult to achieve the desired performance.In order to further improve the accuracy of prediction,we will add multi-view learning to the current network.The model contains two main components: multi-view embedding and inverse-covariance estimation terms.The multi-view embedding term is restricted by low-rank constraints to seek the lowest-rank representation of samples.The inversecovariance estimation term is restricted by structured sparsity regularization to learn a more reasonable distribution prediction model.Experiment results performed on three publicly available datasets demonstrate the effectiveness of our proposed scheme compared with state-of-the-art algorithms.