Download PDFOpen PDF in browserThe Unconscious Response of Multimedia Content Consumers to Emotionally Significant Visual Symbols: Experimental Study and Oculometry DatasetEasyChair Preprint 413710 pages•Date: September 4, 2020AbstractThe publication discusses the problem of forming a dataset that can be used as a training sample for a neural network classifier capable of detection of destructive multimedia content. Special attention is paid to the procedure of experimental re-search, detection of anomalies and obtaining categorical estimates. The obtained dataset includes various emotionally significant visual stimuli and oculometric da-ta of the subjects and makes it possible to use quantitative and qualitative criteria for assigning visual content to one of the classes: «toxic», «indifferent», «sanogenic». The dataset is free for non-commercial use and is available to researchers at Kaggle.com Keyphrases: anomaly detection, destructive multimedia content, neural network classifier, training sample
|