As a new classification platform, deep learning has recently received increasing attention from researchers and has been successfully applied to many domains. In some domains, like bioinformatics and robotics, it is very difficult to construct a large-scale well-annotated dataset due to the expense of data acquisition and costly annotation, which limits its development. Transfer learning relaxes the hypothesis that the training data must be independent and identically distributed (i.i.d.) with the test data, which motivates us to use transfer learning to solve the problem of insufficient training data. This survey focuses on reviewing the current researches of transfer learning by using deep neural network and its applications. We defined deep transfer learning, category and review the recent research works based on the techniques used in deep transfer learning.

Contact electrification between water and a solid surface is crucial for physicochemical processes at water–solid interfaces. However, the nature of the involved processes remains poorly understood, especially in the initial stage of the interface formation. Here we report that H 2 O 2 is spontaneously produced from the hydroxyl groups on the solid surface when contact occurred. The density of hydroxyl groups affects the H 2 O 2 yield. The participation of hydroxyl groups in H 2 O 2 generation is confirmed by mass spectrometric detection of 18 O in the product of the reaction between 4-carboxyphenylboronic acid and 18 O–labeled H 2 O 2 resulting from 18 O 2 plasma treatment of the surface. We propose a model for H 2 O 2 generation based on recombination of the hydroxyl radicals produced from the surface hydroxyl groups in the water–solid contact process. Our observations show that the spontaneous generation of H 2 O 2 is universal on the surfaces of soil and atmospheric fine particles in a humid environment.

Fully crystallized alloys gained by annealing of metallic glasses show excellent rejuvenated catalytic capabilities for ultrafast activation of peroxide. As self-motivated galvanic cells form in the fully crystallized alloys, a grain growth contributing to extensively reduced grain boundaries greatly weakens electron trapping and promotes inner electron transportation, providing a significant insight into exploit novel catalysts.

Developing new functional applications of metallic glasses in catalysis is an active and pivotal topic for materials science as well as novel environmental catalysis processes. Compared to the crystalline materials with highly ordered atomic packing, metallic glass has a simply disordered atomic structure. Recent reports have demonstrated that the metallic glasses are indeed having many superiorly catalytic properties, yet the understanding of the mechanism is insufficient. In this work, the structural relaxation (α-relaxation) by annealing in an amorphous Fe 78 Si 9 B 13 alloy is studied for unraveling the catalytic mechanism at the atomic scale. The volume fractions of the crystalline structures, such as α-Fe, Fe 2 Si, and Fe 2 B, in the as-received and annealed metallic glasses are fully characterized. It is found that the randomly atomic packing structure with weak atomic bonding in the as-received metallic glass has an efficient electron transfer capability, presenting advanced superiorities in the aspects of production rate of hydroxyl radicals (•OH), dye degradation rate (k), and essential degradation ability (K SA ) for water treatment. The discovery of this critically important work unveils why using metallic glasses as catalysts has higher reactivity than the crystalline materials, and more importantly, it provides new research opportunities into the study of synthetic catalysts.

Studies have shown that 2,5-hexanedione (2,5-HD) is the main active metabolite of n-hexane in the human body. The toxicity of n-hexane and 2,5-hexanedione has been extensively researched, but toxicity to the reproductive system, especially the impact on female reproductive function, has been less frequently reported. In this study, we exposed human ovarian granulosa cells to 0, 16, 64, and 256 μM 2,5-HD in vitro for 24 h. Through hematoxylin-eosin (HE) staining, Hoechst 33342 staining, transmission electron microscopy, and flow cytometry using FITC-Annexin V/PI double staining, 2,5-HD was demonstrated to cause significant apoptosis of human ovarian granulosa cells in a dose-dependent manner. As part of our continuing studies, we investigated the underlying apoptosis mechanism of human ovarian granulosa cells exposed to 0, 16, 64, and 256 μM 2,5-HD in vitro for 24 h. Real-time quantitative PCR and Western blot analysis were used to detect changes in the expression of the apoptosis-related BCL-2 family (BCL-2, BAX) and CASPASE family (CASPASE-3) with increasing 2,5-HD concentration. The results showed that with increasing 2,5-HD doses, the expression of BCL-2 decreased. However, a marked dose-dependent increase in the expression of BAX and active CASPASE-3 (p17) was observed in human ovarian granulosa cells. These results suggest that the mechanisms of 2,5-HD causing increased apoptosis in human ovarian granulosa cells might be through BCL-2, BAX, and CASPASE-3 signaling pathways.