Electrochemical metallization (ECM) cell kinetics are strongly determined by the electrolyte and can hardly be altered after the cell has been fabricated. Solid-state property tunable electrolytes in response to external stimuli are therefore desirable to introduce additional operational degree of freedom to the ECM cells, enabling novel applications such as multistate memory and reconfigurable computation. In this work, we use Ge2Sb2Te5(GST) as the electrolyte material whose solid state is switched from the amorphous(a) to the crystalline(c) phase thermally. Electrical heating too is readily achievable. The resistive switching characteristics of the cells with different GST phases are examined. The magnitude of the high resistance, the SET voltage and the on/off ratio are found to be considerably affected by the solid phase of GST, whereas the magnitude of the low resistance is least affected. Moreover, a transition from volatile to nonvolatile SET switching is only observed for c-GST based cell under prolonged voltage sweep, but not for a-GST based cell. This work provides a springboard for more studies on the manipulation of the ECM cell kinetics by tunable electrolyte and the resulting unprecedented device functionalities.

Recommender systems play a crucial role in helping users to find their interested information in various web services such as Amazon, YouTube, and Google News. Various recommender systems, ranging from neighborhood-based, associationrule-based, matrix-factorization-based, to deep learning based, have been developed and deployed in industry. Among them, deep learning based recommender systems become increasingly popular due to their superior performance. In this work, we conduct the first systematic study on data poisoning attacks to deep learning based recommender systems. An attacker's goal is to manipulate a recommender system such that the attacker-chosen target items are recommended to many users. To achieve this goal, our attack injects fake users with carefully crafted ratings to a recommender system. Specifically, we formulate our attack as an optimization problem, such that the injected ratings would maximize the number of normal users to whom the target items are recommended. However, it is challenging to solve the optimization problem because it is a non-convex integer programming problem. To address the challenge, we develop multiple techniques to approximately solve the optimization problem. Our experimental results on three realworld datasets, including small and large datasets, show that our attack is effective and outperforms existing attacks. Moreover, we attempt to detect fake users via statistical analysis of the rating patterns of normal and fake users. Our results show that our attack is still effective and outperforms existing attacks even if such a detector is deployed.

No abstract

IntroductionBrucellosis is a highly prevalent zoonotic disease caused by Brucella spp. Brucella suis S2 vaccination is an effective strategy to prevent animal brucellosis. However, S2 induces antibodies against the smooth lipopolysaccharide,making it challenging to distinguish field infected from vaccinated livestock. Early and accurate diagnosis is essential for infection control and prevention. In this study, we aimed to develop a quick and accurate assay to distinguish the BrucellaS2 vaccine strain from closely related B. abortus and B. melitensis.MethodsWhole-genome sequencing of B. suis S2 was performed, and the sequence was compared with that of the genomes of B. abortus and B. melitensis. One specific gene, GL_0002189, was selected as a marker to differentiate the BrucellaS2vaccine strain from B. abortus and B. melitensis. A loop-mediated isothermal amplification (LAMP) assay was developed, based on the GL_0002189 gene, and then assessed for target specificity, lower limit of detection, and repeatability.ResultsOur results revealed that there was no cross-reaction with other strains, and the LAMP assay displayed high sensitivity for detecting S2 with a minimum detection limit of 18.9×103 copies/µL DNA input, it is nearly 100 times higher than conventional PCR technology. Concordance between the LAMP assay and a conventional polymerase chain reaction method was assessed using 54 blood samples collected from sheep with suspected brucellosis. Total concordance between the two assays was 92.6%, without a significant difference (p > 0.05) in the test results.ConclusionThis is the first report of a LAMP assay for the detection of the B. suis S2vaccine strain. Our approach can be helpful for the control and eradication of brucellosis, and its simplicity in requiring no specialized equipment or personnel makes it useful for implementation in resource-limited settings as well as for field use.

Graph Neural Networks (GNNs) have achieved state-of-the-art performance in various graph structure related tasks such as node classification and graph classification. However, GNNs are vulnerable to adversarial attacks. Existing works mainly focus on attacking GNNs for node classification; nevertheless, the attacks against GNNs for graph classification have not been well explored.In this work, we conduct a systematic study on adversarial attacks against GNNs for graph classification via perturbing the graph structure. In particular, we focus on the most challenging attack, i.e., hard label black-box attack, where an attacker has no knowledge about the target GNN model and can only obtain predicted labels through querying the target model. To achieve this goal, we formulate our attack as an optimization problem, whose objective is to minimize the number of edges to be perturbed in a graph while maintaining the high attack success rate. The original optimization problem is intractable to solve, and we relax the optimization problem to be a tractable one, which is solved with theoretical convergence guarantee. We also design a coarse-grained searching algorithm and a query-efficient gradient computation algorithm to decrease the number of queries to the target GNN model. Our experimental results on three real-world datasets demonstrate that our attack can effectively attack representative GNNs for graph classification with less queries and perturbations. We also evaluate the effectiveness of our attack under two defenses: one is well-designed adversarial graph detector and the other is that the target GNN model itself is equipped with a defense to prevent adversarial graph generation. Our experimental results show that such defenses are not effective enough, which highlights more advanced defenses. CCS CONCEPTS• Security and privacy; • Computing methodologies → Machine learning;