Deep learning (DL) has shown great success in many human-related tasks, which has led to its adoption in many computer vision based applications, such as security surveillance systems, autonomous vehicles and healthcare. Such safety-critical applications have to draw their path to success deployment once they have the capability to overcome safety-critical challenges. Among these challenges are the defense against or/and the detection of the adversarial examples (AEs). Adversaries can carefully craft small, often imperceptible, noise called perturbations to be added to the clean image to generate the AE. The aim of AE is to fool the DL model which makes it a potential risk for DL applications. Many test-time evasion attacks and countermeasures, i.e., defense or detection methods, are proposed in the literature. Moreover, few reviews and surveys were published and theoretically showed the taxonomy of the threats and the countermeasure methods with little focus in AE detection methods. In this paper, we focus on image classification task and attempt to provide a survey for detection methods of test-time evasion attacks on neural network classifiers. A detailed discussion for such methods is provided with experimental results for eight state-of-the-art detectors under different scenarios on four datasets. We also provide potential challenges and future perspectives for this research direction.
There is a pressing need to develop more effective therapeutics to fight cancer. An idyllic chemotherapeutic is expected to overcome drug resistance of tumors and minimize harmful side effects to healthy tissues. Antibody‐functionalized porous silicon nanoparticles loaded with a combination of chemotherapy drug and gold nanoclusters (AuNCs) are developed. These nanocarriers are observed to selectively deliver both payloads, the chemotherapy drug and AuNCs, to human B cells. The accumulation of AuNCs to target cells and subsequent exposure to an external electromagnetic field in the microwave region render them more susceptible to the codelivered drug. This approach represents a targeted two‐stage delivery nanocarrier that benefits from a dual therapeutic action that results in enhanced cytotoxicity. 相似文献
A novel synthetic method to synthesize hydroxyapatite/poly (D,L) lactic acid biocomposite is presented in this study by mixing only the precursors hydroxyapatite and (D,L) LA monomer without adding neither solvent nor catalyst. Three compositions were successfully synthesized with the weight ratios of 1/1, 1/3, and 3/5 (hydroxyapatite/(D,L) lactic acid), and the grafting efficiency of poly (D,L) lactic acid on hydroxyapatite surface reaches up to 84?%. Scanning electron microscopy and Fourier transform infrared spectroscopy showed that the hydroxyapatite particles were successfully incorporated into the poly (D,L) lactic acid polymer and X ray diffraction analysis showed that hydroxyapatite preserved its crystallinity after poly (D,L) lactic acid grafting. Differential scanning calorimetry shows that Tg of hydroxyapatite/poly (D,L) lactic acid composite is less than Tg of pure poly (D,L) lactic acid, which facilitates the shaping of the composite obtained. The addition of poly (D,L) lactic acid improves the adsorption properties of hydroxyapatite for fibronectin extracellular matrix protein. Furthermore, the presence of poly (D,L) lactic acid on hydroxyapatite surface coated with fibronectin enhanced pre-osteoblast STRO-1 adhesion and cell spreading. These results show the promising potential of hydroxyapatite/poly (D,L) lactic acid composite as a bone substitute material for orthopedic applications and bone tissue engineering. 相似文献
The use of external electric and magnetic fields for the synthesis and processing of inorganic materials such as metals and ceramics has seen renewed interest in recent years. Electromagnetic energy can be utilized in different ways to improve or accelerate phase formation and stabilization, chemical ordering, densification and coarsening of particle-based materials (pore elimination and grain growth), and mechanical deformation (plasticity and creep). In these new synthesis and processing routes, the resulting microstructures and macroscopic material behavior are determined by the interaction of the applied fields with defects such as single or clustered point defects, dislocation networks, and interfaces. Multiscale experimental investigations and modeling are necessary to unveil the mechanisms underlying this field-assisted manipulation of matter. 相似文献
Infrared (IR) heating is widely used for thermoforming of thermoplastic polymers. The key benefit of radiation heating is that a significant amount of the radiative energy penetrates into the polymers thanks to their semi-transparency. For the case of heating unfilled semi-crystalline polymers, the relation between their microcrystalline structure and optical properties is the key to develop a predictive IR-heating model as microcrystalline structure introduces an optically heterogeneous medium. In this study, a relation between the microcrystalline structure of a polyethylene (PE) and its effect on the thermo-optical properties was experimentally analyzed considering a two-step analysis. At very first step, the relation was analyzed considering samples with identical thicknesses and different morphologies, characterized here in terms of degree of crystallinity (Xc (%)). Using Fourier Transform Infrared (FT-IR) spectroscopy and integrating sphere, optical characteristics of the PE samples were analyzed in near-infrared (NIR) and middle-infrared (MIR) spectral ranges. The analyses showed that a slight variation in Xc (%) has a great effect on the optical characteristics of PE, particularly the transmission characteristics in NIR range. The wavelength-dependent effect of Xc (%) on the transmission behaviors opened a discussion about the fact that the microcrystalline structures -in particular spherulites or their substructures such as lamellae- are responsible for optical scattering. Using the optical properties obtained from the two-step experimental analyses, two different thermo-optical properties were calculated, namely extinction and absorption coefficients, and used as a numerical input for the parametric numerical studies. The numerical studies were performed using an in-house developed radiation heat transfer algorithm -RAYHEAT-. Both the experimental and numerical analyses demonstrated the importance of the optical scattering regarding the identification of thermo-optical properties, used as a numerical input for radiation heat transfer models. 相似文献
Flash spark plasma sintering (flash SPS) is an attractive method to obtain Nd–Fe–B magnets with anisotropic magnetic properties when starting from melt-spun powders. Compared to the benchmark processing route via hot pressing with subsequent die upsetting, flash SPS promises electroplasticity as an additional deformation mechanism and reduced tool wear, while maximizing magnetic properties by tailoring the microstructure—fully dense and high texture. A detailed parameter study is conducted to understand the influence of Flash SPS parameters on the densification and magnetic properties of commercial MQU-F powder. It is revealed that the presintering conditions and preheating temperature before applying the power pulse play a major role for tailoring grain size and texture in the case of hot deformation via Flash SPS. Detailed microstructure and magnetic domain evaluation disclose the texture enhancement with increasing flash SPS temperature at the expense of coercivity. The best compromise between remanence and coercivity (1.37 T and 1195 kA m−1, respectively) is achieved through a combination of presintering at 500 °C for 120 s and preheating temperature of 600 °C, resulting in a magnet with energy product (BH)max of 350 kJm−3. These findings show the potential of flash SPS to obtain fully dense anisotropic nanocrystalline magnets with high magnetic performance. 相似文献
Novel sintering methods have emerged in the recent past years, which have raised great interest in the scientific community. Relying on electric field effects, high heating rates, the use of mechanical pressure, or hydrothermal conditions, they offer fundamental advantages compared to conventional sintering routes like minimizing the energy consumption and enhancing the process efficiency. This perspective aims at explaining these effects in a general way and presenting the status quo of using them for the processing of high-performing ceramic materials. In detail, this work focuses on flash sintering, ultrafast high-temperature sintering, spark plasma sintering, cold sintering, and photonic sintering methods based on different light sources. The specificities, potentials, and limitations of each method are compared, especially in the light of a possible industrialization. 相似文献
The conversion of low frequency noise into phase noise in microwave oscillators is studied through an analytical calculation of the pushing factor. This calculation is based on a simplified equivalent circuit for two types of active devices : field effect transistors (Fet) and heterojunction bipolar transistors (hbt). The preeminence in the conversion process of the gate- source capacitance in theFet and the base- emitter junction in thehbt is pointed out. Practical methods are proposed to reduce the phase noise in these circuits. 相似文献