Deep learning methods have proven promising performance in decoding specific task states based on functional magnetic resonance imaging (fMRI) of the human brain, however, they lack transparency in their decision making, in the sense that it is not straightforward to visualize the features on which the decision was made. In this study, we investigated the decoding of four sensorimotor tasks based on 3D fMRI according to 3D Convolutional Neural Network (3DCNN), and then adopted Grad-CAM algorithms to provide visual explanation from deep networks so as to support the decoding decision.

No abstract

No abstract

The work aims to study the change rule of the surface properties of aluminum (Al) alloy after shot peening (SP), and obtain the corresponding relationship between the surface material characteristics and SP parameters. Firstly, the Finite Element Model (FEM) of multi-projectile impact Al alloy specimen was established by ANSYS/LS-DYNA (Version: Ansys15.0). Box-Benhnken design method (BBD) was used to design 3-level and 3-factor shot peening experiment with shot peening pressure. Projectile size, jet distance, the surface residual stress, as well as deformation were taken as responses. The surface stress and the deformation at the crater were obtained according to the experiments. Then, Design-Expert software was adopted to fit the values to attain the multiple regression quadratic equations, and the response surface methodology (RSM) was applied to analyse the interaction between the various factors. The degree of model-fitting was simultaneously identified in accordance with an analysis of variance of the function models. Finally, the shot peening test was carried out using 7075-T651 Al alloy as the specimen. Combined with the X-ray diffraction (XRD) stress test and the optical microscopic observation of the crater section, the value of stress and deformation value could be determined to verify the accuracy of the model. The adjusted R2 of the stress function model and the deformation function model were 94.85% and 95.08%, respectively. The deviation between calculated stress value and the experimental value was less than 5.5%. The deformation of the section showed that the deformed layer of the specimen was approximately the same as calculation value. The result indicated the function model had high accuracy. The analysis suggested that the function model could quickly and precisely deduce the parameter combination of the SP from the surface stress or deformation of the material, which provided a diversity reference for the surface stress and hardness strengthening of SP.