Solidification researches using transparent model materials——A review

A review is given in the paper for solidification researches with transparent model materials. The effective experimental me- thod was first proposed by Jackson and Hunt in 1965. The transparent model materials for solidification researches are a kind of non-faceted crystals known as ＂plastic crystals＂ or ＂globular molecules＂, which have very low entropy of melting as that of metals. According to Jackson＇s theory proposed in 1958, entropy of phase transformation will determine whether the phase interface morphology is smooth or rough in atomic scale, which will lead to faceted or nonfacted phase interface in mi- croscopic and macroscopic scales. Succinonitrile （SCN） and its alloys with water, ethanol, acetone, and NH4C1-H：O solution are most commonly used as transparent model materials for solidification researches of dendritic growth, anisotropy of solid-liquid interfacial energy, crystal nucleation, crystal grain formation, directional solidification, eutectic and peritectic so- lidification, solidification defects formation such as bubble, hot tearing, etc. Among these researches, the most impressive work was the critical test of dendritic growth theories with high purity succinonitrile by Glicksman et al., which gave positive answer to the Ivantsov＇s analysis and negative answer to the ad hoc condition of the maximum velocity hypothesis. The future researches with transparent model materials could be suggested in three aspects： 1） accurate measurement of material proper- ties and alloy phase diagrams in more plastic crystals, especially to find more transparent eutectic and peritectic alloys; 2） accurate measurement of the grain boundary groove shape to obtain precise data of the anisotropy parameters of the interfacial free energy in transparent model materials; 3） to get clear pictures of solidification processes with morphology details in a rela- tively large area, with continuous movement of liquid and particles, in order to give experimental support to numerical simula- tions aimming at accurate description of microstructure formation during solidification of multicomponent alloys under complex conditions of real casting and welding processes.

Interpretability of Neural Networks Based on Game-theoretic Interactions

This paper introduces the system of game-theoretic interactions,which connects both the explanation of knowledge encoded in a deep neural networks(DNN)and the explanation of the representation power of a DNN.In this system,we define two gametheoretic interaction indexes,namely the multi-order interaction and the multivariate interaction.More crucially,we use these interaction indexes to explain feature representations encoded in a DNN from the following four aspects:(1)Quantifying knowledge concepts encoded by a DNN;(2)Exploring how a DNN encodes visual concepts,and extracting prototypical concepts encoded in the DNN;(3)Learning optimal baseline values for the Shapley value,and providing a unified perspective to compare fourteen different attribution methods;(4)Theoretically explaining the representation bottleneck of DNNs.Furthermore,we prove the relationship between the interaction encoded in a DNN and the representation power of a DNN(e.g.,generalization power,adversarial transferability,and adversarial robustness).In this way,game-theoretic interactions successfully bridge the gap between“the explanation of knowledge concepts encoded in a DNN”and"the explanation of the representation capacity of a DNN"as a unified explanation.