Component-controlled synthesis of PdxSny nanocrystals on carbon nanotubes as advanced electrocatalysts for oxygen reduction reaction.

Pd-based bimetallic or multimetallic nanocrystals are considered to be potential electrocatalysts for cathodic oxygen reduction reaction (ORR) in fuel cells. Although much advance has been made, the synthesis of component-controlled Pd-Sn alloy nanocrystals or corresponding nanohybrids is still challenging, and the electrocatalytic ORR properties are not fully explored. Herein, component-controlled synthesis of PdxSny nanocrystals (including Pd3Sn, Pd2Sn, Pd3Sn2, and PdSn) has been realized, which are in situ grown or deposited on pre-treated multi-walled carbon nanotubes (CNTs) to form well-coupled nanohybrids (NHs) by a facile one-pot non-hydrolytic system thermolysis method. In alkaline media, all the resultant PdxSny/CNTs NHs are effective at catalyzing ORR. Among them, the Pd3Sn/CNTs NHs exhibit the best catalytic activity with the half-wave potential of 0.85 V (vs. RHE), good cyclic stability, and excellent methanol-tolerant capability due to the suited Pd-Sn alloy component and its strong interaction or efficient electronic coupling with CNTs. This work is conducive to the advancement of Pd-based nanoalloy catalysts by combining component engineering and a hybridization strategy and promoting their application in clean energy devices.

A Weighted GraphSAGE-Based Context-Aware Approach for Big Data Access Control.

Context information is the key element to realizing dynamic access control of big data. However, existing context-aware access control (CAAC) methods do not support automatic context awareness and cannot automatically model and reason about context relationships. To solve these problems, this article proposes a weighted GraphSAGE-based context-aware approach for big data access control. First, graph modeling is performed on the access record data set and transforms the access control context-awareness problem into a graph neural network (GNN) node learning problem. Then, a GNN model WGraphSAGE is proposed to achieve automatic context awareness and automatic generation of CAAC rules. Finally, weighted neighbor sampling and weighted aggregation algorithms are designed for the model to realize automatic modeling and reasoning of node relationships and relationship strengths simultaneously in the graph node learning process. The experiment results show that the proposed method has obvious advantages in context awareness and context relationship reasoning compared with similar GNN models. Meanwhile, it obtains better results in dynamic access control decisions than the existing CAAC models.

Adversarial robustness in deep neural networks based on variable attributes of the stochastic ensemble model.

Deep neural networks (DNNs) have been shown to be susceptible to critical vulnerabilities when attacked by adversarial samples. This has prompted the development of attack and defense strategies similar to those used in cyberspace security. The dependence of such strategies on attack and defense mechanisms makes the associated algorithms on both sides appear as closely processes, with the defense method being particularly passive in these processes. Inspired by the dynamic defense approach proposed in cyberspace to address endless arm races, this article defines ensemble quantity, network structure, and smoothing parameters as variable ensemble attributes and proposes a stochastic ensemble strategy based on heterogeneous and redundant sub-models. The proposed method introduces the diversity and randomness characteristic of deep neural networks to alter the fixed correspondence gradient between input and output. The unpredictability and diversity of the gradients make it more difficult for attackers to directly implement white-box attacks, helping to address the extreme transferability and vulnerability of ensemble models under white-box attacks. Experimental comparison of ASR-vs.-distortion curves with different attack scenarios under CIFAR10 preliminarily demonstrates the effectiveness of the proposed method that even the highest-capacity attacker cannot easily outperform the attack success rate associated with the ensemble smoothed model, especially for untargeted attacks.