Fast generalization error bound of deep learning without scale invariance of activation functions.

In the theoretical analysis of deep learning, discovering which features of deep learning lead to good performance is an important task. Using the framework for analyzing the generalization error developed by Suzuki (2018), we derive a fast learning rate for deep neural networks with general activation functions. According to Suzuki (2018), scale invariance of the activation functions is essential to derive tight error bounds.

Upregulated expression of FGF13/FHF2 mediates resistance to platinum drugs in cervical cancer cells.

Cancer cells often develop drug resistance. In cisplatin-resistant HeLa cisR cells, fibroblast growth factor 13 (FGF13/FHF2) gene and protein expression was strongly upregulated, and intracellular platinum concentrations were kept low. When the FGF13 expression was suppressed, both the cells' resistance to platinum drugs and their ability to keep intracellular platinum low were abolished.

Graded-index plastic optical fiber prepared by the coextrusion process.

A graded-index plastic optical fiber (GI POF) has been proposed as the transmission medium to realize high-speed information transmission. We have succeeded in fabricating a GI POF by the dopant diffusion coextrusion process, a method that allows continuous fabrication of GI POF. Although it has been indicated that the refractive index distribution of GI POF fabricated by this process is formed by Fick diffusion with the diffusion coefficient dependent on dopant concentration, the method to control it remains unknown.