Graph convolutional neural networks have recently shown great potential for the task of zero-shot learning. These models are highly sample efficient as related concepts in the graph structure share statistical strength allowing generalization to new classes when faced with a lack of data. However, multi-layer architectures, which are required to propagate knowledge to distant nodes in the graph, dilute the knowledge by performing extensive Laplacian smoothing at each layer and thereby consequently decrease performance. In order to still enjoy the benefit brought by the graph structure while preventing dilution of knowledge from distant nodes, we propose a Dense Graph Propagation (DGP) module with carefully designed direct links among distant nodes. DGP allows us to exploit the hierarchical graph structure of the knowledge graph through additional connections. These connections are added based on a node's relationship to its ancestors and descendants. A weighting scheme is further used to weigh their contribution depending on the distance to the node to improve information propagation in the graph. Combined with finetuning of the representations in a two-stage training approach our method outperforms state-of-the-art zero-shot learning approaches.

Encouraged by the increasing requirements of intelligent equipment, silicon integrated circuit–compatible photodetectors that support single‐chip photonic–electronic systems have gained considerable progresses. Advanced materials have resulted in enhanced device performance based on traditional photovoltaic effect and photoconductive effect, and novel device designs have catalyzed new working mechanisms combing rapid photoresponse and high responsivity gain. Surprising applications are developed using monolithic photonic–electronic platforms, and the developing integration strategies keep pace with the developing complementary metal‐oxide‐semiconductor techniques as well as nonsilicon substrates. Here, the recent developments in silicon‐compatible photodetectors, both in device advances and their integration routes, are reviewed. Meanwhile, the progresses, challenges, and possible future directions in this field are discussed and concluded.

The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adma.201805722.Self-powered electronics using triboelectric nanogenerators (TENGs) is drawing increasing efforts and rapid advancements in eco/biocompatible energy harvesting, intelligent sensing, and biomedical applications. Currently, the triboelectric performances are mainly determined by the pair materials' inherent electron affinity difference, and merely tuned by chemical or physical methods, which significantly limit the optional variety and output capability, especially for natural-biomaterial-based TENGs. Herein, a biocompatible triboelectric material with a programmable triboelectric property, multiple functionalization, large-scale-fabrication capability, and transcendent output performance is designed, by genetically engineering recombinant spider silk proteins (RSSP). Featuring totally "green" large-scale manufacturing, the water lithography technique is introduced to the RSSP-TENG with facilely adjustable surface morphology, chemically modifiable surface properties, and controllable protein conformation. By virtue of the high electrical power, a proof-of-principle drug-free RSSP-patch is built, showing outstanding antibacterial performances both in vitro and in vivo. This work provides a novel high-performance biomaterial-based TENG and extends its potential for multifunctional applications.