Abstract: A method for assessing test adequacy of deep neural networks based on element decomposition is provided. The network testing is divided into black box testing and white box testing, of which key elements are decomposed and defined. Network parameters including a weight matrix and a bias vector are extracted. Importance values of neurons in individual layers of the deep neural network are calculated and clustered, and an importance value hot map of neurons in each layer is generated based on clustering results. Mutation testing, and index calculation and evaluation are performed.

Abstract: The invention provides a drying apparatus and a method for a raw silk degumming process. The drying apparatus includes a drying oven in which a silk strand cover and an air supply unit are disposed. An upper opening and a lower opening are respectively provided at the top and the bottom of the silk strand cover. A silk strand to be dried passes through the silk strand cover. A convex portion is disposed on a sidewall of the silk strand cover, which is protruded toward an outer side of the silk strand cover. The air supply unit is used for providing an air supply flow into the silk strand cover to dry the silk strand. A falling point of the air supply flow is located at the convex portion of the silk strand cover. By means of the apparatus, silkworm silk can be endowed with special elasticity and fluffiness performances.