Alpha Fusion Adversarial Attack Analysis Using Deep Learning

The deep learning model encompasses a powerful learning ability that integrates the feature extraction, and classification method to improve accuracy. Convolutional Neural Networks (CNN) perform well in machine learning and image processing tasks like segmentation, classification, detection, identification, etc. The CNN models are still sensitive to noise and attack. The smallest change in training images as in an adversarial attack can greatly decrease the accuracy of the CNN model. This paper presents an alpha fusion attack analysis and generates defense against adversarial attacks. The proposed work is divided into three phases: firstly, an MLSTM-based CNN classification model is developed for classifying COVID-CT images. Secondly, an alpha fusion attack is generated to fool the classification model. The alpha fusion attack is tested in the last phase on a modified LSTM-based CNN (CNN-MLSTM) model and other pre-trained models. The results of CNN models show that the accuracy of these models dropped greatly after the alpha-fusion attack. The highest F1 score before the attack was achieved is 97.45 And after the attack lowest F1 score recorded is 22%. Results elucidate the performance in terms of accuracy, precision, F1 score and Recall.     

The mechanical properties of epoxy resins

Crack propagation in a series of epoxy resins described in Part 1 has been studied as a function of testing rate and temperature. It has been found that crack propagation is continuous at low temperatures but that as the temperature is raised the mode of propagation becomes unstable (stick/slip). Features on the fracture surfaces at the crack arrest lines have been shown to be of the same dimensions as those expected for a Dugdale plastic zone. It has been suggested that the slip process takes place by slow growth of a crack through the plastic zone followed by rapid propagation through virgin material. It has been shown that the stick/slip behaviour is due to blunting of the crack which is controlled by the yield behaviour of the resin. A unique fracture criterion has been shown to be applicable to epoxy resins which is that a critical stress of the order of three times the yield stress must be achieved at a critical distance ahead of the crack. Electron microscope replicas of the fracture surfaces have been obtained and an underlying nodular structure can be resolved. However, no direct correlation between the nodule size and fracture properties has been found.     

Crack propagation in and fractography of epoxy resins

The relationship between the stability of crack propagation in, and the fracture surface appearance of, DGEBA epoxy resins cured with TETA has been investigated using a linear elastic fracture mechanics approach. In particular, the effect of varying the amount of curing agent and curing conditions and altering external variables such as testing rate, temperature and environment have been studied. Under certain conditions, propagation is found to be stable and fracture surfaces have a smooth appearance. Under other conditions the cracks propagate in an unstable stick-slip manner. In this case, arrest lines can be seen on the fracture surfaces and their intensity and roughness increases with the magnitude of the crack jumps. The roughness of the fracture surfaces has been measured using a Talysurf and this has been shown to be due principally to deviation of the cracks from the original fracture plane rather than any gross plastic deformation.     

Human SMILE-Derived Stromal Lenticule Scaffold for Regenerative Therapy: Review and Perspectives

A transparent cornea is paramount for vision. Corneal opacity is one of the leading causes of blindness. Although conventional corneal transplantation has been successful in recovering patients’ vision, the outcomes are challenged by a global lack of donor tissue availability. Bioengineered corneal tissues are gaining momentum as a new source for corneal wound healing and scar management. Extracellular matrix (ECM)-scaffold-based engineering offers a new perspective on corneal regenerative medicine. Ultrathin stromal laminar tissues obtained from lenticule-based refractive correction procedures, such as SMall Incision Lenticule Extraction (SMILE), are an accessible and novel source of collagen-rich ECM scaffolds with high mechanical strength, biocompatibility, and transparency. After customization (including decellularization), these lenticules can serve as an acellular scaffold niche to repopulate cells, including stromal keratocytes and stem cells, with functional phenotypes. The intrastromal transplantation of these cell/tissue composites can regenerate native-like corneal stromal tissue and restore corneal transparency. This review highlights the current status of ECM-scaffold-based engineering with cells, along with the development of drug and growth factor delivery systems, and elucidates the potential uses of stromal lenticule scaffolds in regenerative therapeutics.     

Enzymatic Synthesis of Glucose- and Xylose Laurate Esters Using Different Acyl Donors,Higher Substrate Concentrations,and Membrane Assisted Solvent Recovery

Glucose- and xylose laurate esters are enzymatically synthesized using equimolar substrate concentrations in 2-methyl-2-butanol, comparing free lauric acid with methyl- and vinyl-laurate as acyl donors. All reactions result in ≥70% acyl donor conversions after 72 h but the activated donors are also partially hydrolyzed to lauric acid, highlighting the difficulty in controlling water presence in this particular reaction system. The esterification of xylose generates a complex product profile, with several regioisomers of monoesters and diesters. The esterification of glucose is quite selective, forming mainly the 6-O monoester (≥96%) with a small presence of two diester isomers (4%). Increasing substrate concentration up to 800 millimoles kg⁻¹ results in lower conversion values (down to 58%) but shows that the reaction proceeds successfully even in the presence of high amounts of insoluble glucose. However, the reaction is less selective and the proportion of diester increases, becoming up to 46% (molar fraction) of the final product. Solvent recovery after esterification can be achieved by organic solvent nanofiltration through a polymeric membrane able to retain ≥80% of all reaction substrates and products. Practical Applications: The use of high substrate concentrations during the enzymatic synthesis of sugar ester biosurfactants leads to product titers that are more industrially appealing, without the need to find a solvent that can solubilize all initial substrate. The sustainability of the enzymatic conversion at mild temperatures can be enhanced by recycling of the reaction solvent through organic solvent nanofiltration, an energy efficient alternative to other traditional methods like distillation.     

Combination of supercritical fluid extraction with dispersive liquid-liquid microextraction for extraction of organophosphorus pesticides from soil and marine sediment samples

Supercritical fluid extraction (SFE) coupled with dispersive liquid-liquid microextraction (DLLME) and followed by gas chromatography-flame ionization detection (GC-FID) was applied for extraction and determination of ultra-trace amounts of seven organophosphorus pesticides (OPPs) (o,o,o-triethyl phosphorothioate, thionazin, sulfotepp, disulfoton, methyl parathion, parathion, and famphur) in soil and marine sediment samples. Supercritical CO₂ at 150 bar, 60 °C, 10 min static and 30 min dynamic extraction times was used to extract the pesticides. The extracts were collected in 1.0 mL of acetonitrile. Seventeen μL of carbon tetrachloride was dissolved in the collecting solvent and the mixture was then injected rapidly into 5.0 mL of aqueous solution. About (5.0 ± 0.2 μL) of sediment phase was collected after centrifuging and finally 2.0 μL of it was injected into gas chromatography (GC) injection port for analyses. The extraction recoveries for the target analytes were in the range of 44.4% and 95.4% and relative standard deviation (RSD%) for four-replicate measurements was below 7.5%. The limit detections of the method for determining the pesticides were in the range of 0.001-0.009 mg kg⁻¹. The method was successfully applied for analysis of OPPs in real soil and marine sediment samples and satisfactory results were obtained.     

Comparison of essential oils compositions of Nepeta persica obtained by supercritical carbon dioxide extraction and steam distillation methods

Essential oil of Nepeta persica cultivated in Iran was obtained by steam distillation and supercritical (carbon dioxide) extraction methods. The oils were analysed by capillary gas chromatography using flame ionization and mass spectrometric detections. The compounds were identified according to their retention indices and mass spectra (EI, 70 eV). The effects of different parameters such as pressure, temperature, modifier volume and extraction times (dynamic and static) on the supercritical fluid extraction (SFE) of N. persica oil were investigated. The results showed that under the pressure of 20.3 MPa, temperature of 45 °C, methanol of 1.5% v/v), dynamic extraction time of 50 min and static extraction time of 25 min extraction was more selective for the 4αβ,7α,7aα-nepetalactone. Twelve compounds were identified in the steam-distilled oil. The major components of N. persica were 4αβ,7α,7aα-nepetalactone (26.5%), cis-β-farnesene (4.4%) and 3,4α-dihydro-4aα,7α,7aα-nepetalactone (3.5%). However, by using supercritical carbon dioxide under optimum conditions, only two components have more than 90.0% of the oil. The extraction yield based on steam distillation was 0.08% (v/w). On the other hand, using SFE extraction yield in the range of 0.22–8.90% (w/w) were obtained at different conditions. The results show that, in Iranian N. persica oil, 4αβ,7α,7aα-nepetalactone is a major component.