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: Disclosed are a heat sink base and a heat sink. The heat sink base includes a base body. The bottom of the base body is provided with at least one protruded structure, and the bottom of the base body abuts against the top of the CPU. Each of the at least one protruded structure includes a gentle region with a protrusion, and a slope region surrounding the protruded gentle region with the protrusion. The gentle region of the protruded structure abuts against a high heat flux region at the top of the CPU, thus forming a region with relatively high pressure and relatively stable pressure compared with the slope region. Therefore, the heat sink base provided by the present invention has a simple and reasonable structure, and can efficiently dissipate the heat generated by the CPU, thus improving the cooling efficiency and obtaining more stable and more consistent cooling performance.