Search Patent of the week: Classifying resources using a deep network
This episode discusses a patent that describes a system and method for categorizing resources using a deep neural network. Resource features are converted into numerical values by embedding functions, processed in neural network layers and finally evaluated by a classifier to predict category membership. The method is used in search engines, for example to detect spam and optimize search results. The patent covers various implementations of the system and method, including software and hardware aspects. The accuracy and efficiency of the categorization are emphasized. https://www.kopp-online-marketing.com/patents-papers/classifying-resources-using-a-deep-network