Uniforming the dimensionality of data with neural networks for materials informatics

• We introduced two representation in respect to materials: naive representation and expert representation.
• We proposed a uniforming method of dimensionality of data with different size input vectors using a neural network.
• The proposed method outperformed the conventional methods (the multi-layer autoencoder, the denoising autoencoder, and kernel PCA) for the linear regression task on synthetic data.
• The experimental results showed the robustness for data size and number of constituent elements.
• In the linear regression task on ion conductivity data of bulk materials and hydrogen storage materials, the good fitting performance was obtained in terms of the latent data uniformed by the proposed method.

Materials informatics is a growing field in materials science. Materials scientists have begun to use soft computing techniques to discover novel materials. In order to apply these techniques, the descriptors (referred to as features in computer science) of a material must be selected, thereby deciding the resulting performance. As a way of describing a material, the properties of each element in the material are used directly as the features of the input variable. Depending on the number of elements in the material, the dimensionality of the input may differ. Hence, it is not possible to apply the same model to materials with different numbers of elements for tasks such as regression or discrimination. In the present paper, we present a novel method of uniforming the dimensionality of the input that allows regression or discriminative tasks to be performed using soft computing techniques. The main contribution of the proposed method is to provide a solution for uniforming the dimensionality among input vectors of different size. The proposed method is a variant of the denoising autoencoder Vincent et al. (2008) [1] using neural networks and gives a latent representation with uniformed dimensionality of the input. In the experiments of the present study, we consider compounds with ionic conductivity and hydrogen storage materials. The results of the experiments indicate that the regression tasks can be performed using the uniformed latent data learned by the proposed method. Moreover, in the clustering task using these latent data, we observed distance preservation in data space, which is also the case for the denoising autoencoder. This result may enable the proposed method to be used in a broad range of applications.

Figure optionsDownload as PowerPoint slide