Dense Layer Research Articles

Oxynitrided Ti-6Al-4V Coatings Deposited by Twin Wire Arc Spray for Protection of Aluminum Die-Casting Molds

Abstract Steel molds used for aluminum die-casting often fail due to excessive wear or cracking phenomena associated with the soldering effect in contact with molten aluminum, which leads to the formation of iron-based intermetallic compounds and causes problems in the cast components. One solution is to apply protective coatings whose composition is less reactive with the molten aluminum and improve its hardness, toughness, wear, and corrosion resistance, thus prolonging its service life. This work evaluates the effectiveness of Ti-6Al-4V coatings deposited by twin wire arc spraying in an air or nitrogen atmosphere. Nitrogen was used as the carrier and shielding gas for the in-flight molten particles. Coatings were deposited by varying the stand-off distance, the nitrogen gas pressures, and the substrate temperature. The microstructure of the coatings is interlayered, one porous layer of dendrites and one highly densified layer. The presence of TiN, TiO2, α-Ti, and β-Ti phases was confirmed by different characterization methods. For instance, x-ray photoelectron spectroscopy measurements confirmed the presence of N-Ti, O-Ti, and N-O bonds, with the oxygen/nitrogen/titanium percentage associated with the formation of a non-stoichiometric (Ti, Al, V)NxOy phase. Finally, the reactivity of selected oxynitrided Ti-6Al-4V coating in contact with molten aluminum showed a low reaction rate compared to the coarse reaction layer suffered by the uncoated steel substrates.

A Novel Delayed Phase Inversion Strategy Enables Green PVDF Membranes for Membrane Distillation

Polyvinylidene fluoride (PVDF) membranes are extensively utilized in membrane distillation (MD) for water treatment. However, traditional methods easily form asymmetrical membranes with dense skin layers that are detrimental to membrane flux. Herein, an eco-friendly PVDF membrane was fabricated by utilizing a delayed phase separation process without using any pore-forming agents. In addition, methyl-5-(dimethylamino)-2-methyl-5-oxopentanoate (PolarClean) was used as a green solvent without posing risks to humans and the environment. It was demonstrated that the PVDF concentration is crucial in influencing the microstructures and performance of the resulting membranes. As the PVDF concentration increased, the morphology changed significantly, resulting in a reduction of pore size. When feeding the device with NaCl solution at a concentration of 35 g/L, the MD water vapor flux reached 18.49 kg·m⁻2·h⁻1, while maintaining a salt rejection of over 99.97% during the continuous operation for 24 h. This work presented a method for producing green PVDF membranes via delayed phase inversion with satisfactory water vapor flux and salt rejection, highlighting their prospect for effective applications in MD for water treatment.

Effects of Cleaning Interval, Hapa Layers and Stocking Density on Growth Performance of Sandfish Holothuria scabra Juveniles in Ocean‐Based Nursery Production Systems

ABSTRACTThe mass production of release‐size H. scabra juveniles in the Philippines is primarily done in a floating hapa ocean‐based nursery systems. To optimize its production, understanding the influence of different culture conditions and their management is necessary. In this study, the effects of hapa net cleaning interval, hapa layers, and stocking density on growth and survival of sandfish juveniles reared in floating hapa (L × W × H: 1.7 m × 0.9 m × 1.2 m) nursery system were assessed in two 60‐day field studies. Results showed that the number of hapa net layers and cleaning interval had negligible effects on the growth performance of sandfish juveniles by day 30. However, sandfish juvenile survival was affected by hapa net layers with higher survival in triple‐layer mosquito net hapas (40.5 ± 0.97%) compared to double‐layer mosquito net hapas (17.8 ± 0.62%). By day 60, the growth performance of sandfish juveniles in double‐layer hapas was higher (average weight: 5.0 ± 0.97 g; AGR: 0.13 ± 0.03 g day−1) compared to that in triple‐layer mosquito hapa nets (average weight: 3.1 ± 0.23 g; AGR: 0.07 ± 0.01 g day−1). Survival of sandfish juveniles reared in double‐layer mosquito net hapas, cleaned at 5‐day intervals (85.9 ± 4.58%) was significantly lower compared to all other treatments. Conversely, initial juvenile stocking density did not significantly affect the growth performance and survival of sandfish juveniles in both rearing cycles. We therefore recommend the use of triple‐layer mosquito net hapas during the first 30‐day culture period and an initial stocking density of 1000 post‐settled sandfish juveniles to increase production output. After 30 days, when the juveniles are larger, the use of double‐layer mosquito net hapas and lower stocking density are recommended to reduce cost and enhance growth performance. Where cleaning is necessary, hapa nets may be cleaned at 10‐day intervals to improve survival. The results of this study have important implications for optimizing the production of release‐size sandfish in ocean‐based nursery systems in the Philippines.

Robust Multi-Class Classification for Real-Time Agricultural Applications Using Efficient and Adaptive Deep Learning

Plant diseases severely affect agricultural productivity, necessitating accurate and rapid detection methods. This research presents a robust, multi-class plant disease classification framework using adaptive deep learning. We utilize pre-trained convolutional neural networks (CNNs), specifically Xception, InceptionResNetV2, InceptionV3, ResNet50, and the proposed EfficientNetB3-based Adaptive Augmented Deep Learning (EfficientNetB3-AADL) model. Our approach leverages transfer learning combined with extensive data augmentation and trimming techniques to enhance model performance and mitigate overfitting. The EfficientNetB3-AADL architecture incorporates convolutional and max pooling layers, regularization strategies, and a dense feature learning layer, optimized to classify 52 disease categories from a publicly available leaf image dataset. The model’s performance is extensively evaluated using metrics such as accuracy, precision, recall, and F1 score. Notably, EfficientNetB3-AADL achieves superior accuracy over 98%, outperforming other CNN models. The proposed methodology highlights the efficacy of compound scaling and adaptive data augmentation in ensuring robust and efficient disease classification, suitable for real-time agricultural applications. This advancement supports sustainable farming by offering a scalable, computationally efficient solution for early and accurate disease detection in diverse crop species.

An Analysis of the Conceptual and Functional Factors Affecting the Effectiveness of Proton-Exchange Membrane Water Electrolysis

Hydrogen has the potential to decarbonize the energy and industrial sectors in the future, mainly if it is generated by water electrolysis. The proton-exchange membrane water electrolysis (PEMWE) system is regarded as a propitious technology to produce green hydrogen from water using power supplied by renewable energy sources. It offers many benefits, such as high performance, high proton conductibility, quick response, compact size, and low working temperature. Many conceptual and functional parameters influence the effectiveness of PEM, including temperature, pressure of anode and cathode regions, water content and wideness of the layer, and cathode and anode exchange current density. In addition, the anodic half-reaction (known as the oxygen evolution reaction (OER)) and cathodic half-reaction (known as the hydrogen evolution reaction (HER)) perform an important function in the development of PEMWE. The current study aims to present these parameters and discuss their impacts on the performance of PEM. Also, the PEM efficiency is presented. The different methods used to enhance the scattering of OER electrocatalysts and minimize catalyst loading to minimize the price of PEMWE are also highlighted. Moreover, the alternative noble metals that could be used as electrocatalysts in HER and OER to minimize the cost of PEM are reviewed and presented.

Review of hymenolepidids associated with the tribe Sudarikovinini Spassky, 1991 (Eucestoda, Hymenolepididae), with descriptions of two unrecognized species from the Shaw's jird Meriones (Pallasiomys) shawi (Duvernoy) (Rodentia: Muridae: Gerbillinae) from Northern Africa.

The taxonomic position of the hymenolepidid cestode genera Paraoligorchis Wason et Johnson, 1977 and Sudarikovina Spassky, 1951 from gerbils and ground squirrels from Africa, Southwest Asia and South Asia is clarified based on re-examination of type and voucher materials. The generic diagnosis of Sudarikovina is amended. The main corrections for diagnostic characters at the generic level include: irregularly-spaced transverse anastomoses of ventral osmoregulatory canals; presence of both specimens with dextral genital pores and specimens with sinistral genital pores; vagina clearly covered externally by a dense layer of intensely-stained cells; young uterus reticulate and fully-developed uterus with a labyrinthine structure. The previously proposed generic character of a minimum number of testes per proglottis of species of Sudarikovina is reduced to 3. The genus Paraoligorchis is recognized as a junior synonym of Sudarikovina. For the type species of Paraoligorchis, P. taterae Wason et Johnson, 1977, the replacement name Sudarikovina wasojohni nom. nov. is proposed, thus resolving the secondary homonymy of the transferred species and its senior homonym S. taterae Hunkeler, 1972. The validity of the tribe Sudarikovinini sensu Spassky (1991), to which Paraoligorchis and Sudarikovina had been attributed, cannot be resolved due the lack of molecular data. The cestode specimens from Shaw's jird, Meriones (Pallasiomys) shawi (Duvernoy) from Algeria and Tunisia, initially designated as Hymenolepis sp. by Joyeux (1923) and Hymenolepis procera Janicki, 1904 by Joyeux & Foley (1930), represent undescribed species of Sudarikovina, which are described in this study as Sudarikovina pentatesticulata sp. n. and S. tetratesticulata sp. n., respectively. The main diagnostic characters of the two new species, which distinguish them from their congeners, are the number and the size of the testes.

Abstract 4136597: Atherosclerotic plaque healing as a marker of plaque vulnerability: Insight from intracoronary polarization-sensitive optical frequency domain imaging

Background: Collagen is the major extracellular matrix that imparts mechanical strength to fibrous caps covering fibroatheromas. Intracoronary polarimetry with polarization-sensitive (PS) optical frequency domain imaging (OFDI) measures polarization properties, such as birefringence and depolarization (Figure 1). Birefringence is elevated in collagen and collagen-synthesizing smooth muscle cells, while depolarization is increased by the presence of macrophages and lipid/necrotic cores. Purpose: This study aimed to investigate polarimetric signatures of coronary lesions with healed coronary plaques (HCP) in patients with ACS and chronic coronary syndrome (CCS). Furthermore, we aimed to investigate diagnostic value for birefringence and depolarization of ACS culprit lesions discriminating from CCS target lesions. Methods: We conducted a single center prospective registry of intracoronary PS-OFDI imaging in patients with coronary artery disease (n = 50). A total of 862 OFDI frames selected from culprit or target lesions were analyzed. Coronary plaque phenotypes were assessed using conventional OFDI imaging. HCP was defined as plaques with one or more layers of different optical density and a clear demarcation from underlying components on intensity images. Birefringence and depolarization of the newer intima was measured in cross-sectional images. Birefringence and depolarization of ACS- and CCS-lesions were compared using a generalized estimating equation model. Receiver operating characteristic (ROC) analysis was used to investigate the diagnostic performance of polarimetric signatures for identifying ACS-lesions. Results: Compared to CCS-lesions, ACS-lesions featured significantly higher lipid-burden index and maximum lipid arc (both p < 0.05). Compared to the CCS-lesions, ACS-lesions exhibited significantly lower birefringence (p < 0.05) and higher depolarization (p < 0.05). In the ROC analysis for differentiating ACS-lesions from CCS-lesions, area under the curves (AUC) for birefringence and depolarization were 0.712 and 0.672, respectively. In the multivariable ROC analysis in diagnosing ACS lesions, combination of birefringence with depolarization improved the AUC to 0.755 (p ＝ 0.025). Conclusions: Intracoronary polarimetry provides quantitative assessment of plaque composition in patients. Further research is warranted to investigate whether birefringence can serve as a marker of healing failure following plaque rupture and erosion.

Interlayer-Functionalized Graphene with Phosphorus–Silicon-Containing Elements for Improving Thermal Stability and Flame Retardance of Polyacrylonitrile

Highly-effective non-halogenated flame retardants have received widespread attention because they are environmentally friendly, with low toxicity and low smoke density. In this work, interlayer-functionalized graphene (fRGO) containing silicon and phosphorus elements was synthesized via hydrolytic condensation with 3-(methacryloyloxy)propyltrimethoxysilane and addition reaction with 9,10-dihydro-9-oxa-10-phosphaphenanthrene-10-oxide. Interlayer spacing and oxygen-containing groups of reduced graphene oxide (RGO) were regulated by controlling the hydrazine hydrate dosage. Then, phosphorus–silicon-containing organic molecules were inserted into RGO interlayers; this was verified by FTIR, XPS, TEM, etc. The fRGO was added to a polyacrylonitrile (PAN) matrix using a solution blending method to prepare polyacrylonitrile (PAN) composites. The fRGO addition caused the significant decrease in cyclization heat and the considerable increase in char residues, indicating improved thermal stability. Importantly, PAN composites exhibited outstanding flame-retardant properties, with the peak heat release rate reduced by 45%, which is ascribed to the dense graphitic carbon layers induced by phosphorus–silicon-containing organics and the 2D barrier effect of RGO layers to prevent the heat and mass transfer.

Experimental investigation of circumnutation-inspired penetration in sand

Probes that penetrate soil are used in fields such as geotechnical engineering, agriculture, and ecology to classify soils and characterize their propertiesin situ. Conventional tools such as the Cone Penetration Test (CPT) often face challenges due to the lack of reaction force needed to penetrate stiff or dense soil layers, necessitating the use of large drill rigs. This paper investigates more efficient means of penetrating soil by taking inspiration from a plant-root motion known as circumnutation. Experimental penetration tests on sands are performed with circumnutation-inspired (CI) probes that advance at a constant vertical velocity (v) while simultaneously rotating at a constant angular velocity (ω). These probes have bent tips with a given bent angle (α) and bent length (L1). The variation of the mobilized vertical force (Fz), torque (Tz.), and the mechanical work components with the ratio of tangential to vertical velocity (ωR/ν, whereRis the distance of the tip of the probe from the vertical axis of rotation) is investigated along with the effects of probe geometry, vertical velocity, and soil relative density (DR). The results show that the soil penetration resistance does not vary withv, but it increases asα,L1, andDRare increased.Fzdecays exponentially with increasingωR/v,Tzinitially increases and then plateaus, while total work (WT) shows little magnitude changes initially but later increases monotonically. The mechanisms leading to these trends are identified as the changes in the probe projected areas and mobilized normal stresses due to differences in probe geometry and the effects ofωR/von the resultant force direction and soil disturbance. The results show that CI penetration within a specific range ofωR/vleads to small increases inWT(i.e.,⩽25%), yet mobilizesFzmagnitudes that are 50%-80% lower than that mobilized during non-rotational penetration (i.e., CPT). This indicates that CI penetration can be adopted forin situcharacterization or sensor placement with smaller vertical forces, allowing for use of lighter rigs.

A Convolution Auto-Encoders Network for Aero-Engine Hot Jet FT-IR Spectrum Feature Extraction and Classification

Aiming at classification and recognition of aero-engines, two telemetry Fourier transform infrared (FT-IR) spectrometers are utilized to measure the infrared spectrum of the areo-engine hot jet, meanwhile a spectrum dataset of six types of areo-engines is established. In this paper, a convolutional autoencoder (CAE) is designed for spectral feature extraction and classification, which is composed of coding network, decoding network, and classification network. The encoder network consists of convolutional layers and maximum pooling layers, the decoder network consists of up-sampling layers and deconvolution layers, and the classification network consists of a flattened layer and a dense layer. In the experiment, data for the spectral dataset were randomly sampled at a ratio of 8:1:1 to produce the training set, validation set, and prediction set, and the performance measures were accuracy, precision, recall, confusion matrix, F1 score, ROC curve, and AUC value. The experimental result of CAE reached 96% accuracy and the prediction running time was 1.57 s. Compared with the classical PCA feature extraction and SVM, XGBoost, AdaBoost, and Random Forest classifier algorithms, as well as AE, CSAE, and CVAE deep learning classification methods, the CAE network can achieve higher accuracy and efficiency and can complete the spectral classification task.

Orientation-dependent strain and dislocation in HVPE-grown α -Ga2O3 epilayers on sapphire substrates

Orientation-dependent substrates provide effective platforms for achieving α-Ga2O3 with low dislocation densities, whereas the associated strain and dislocation dynamics have not been fully explored. Herein, we investigated the evolution of growth mode, interfacial strain, and dislocation propagation in the α-Ga2O3 epitaxial layer with various orientations, grown by the halide vapor-phase epitaxy. Strain tensor theory and geometric phase analysis indicate that the m-plane α-Ga2O3 epitaxial layer exhibits the lowest misfit tensile strain, measured at εxx = 1.46% and εyy = 1.81%, resulting in the lowest edge dislocation density. The m-plane lattice exhibits an inclination of 33.60°, while the c-plane lattice is horizontally aligned and the a-plane lattice oriented perpendicularly. The orientation-dependent growth significantly influences stress relaxation through the generation of misfit dislocations, originating from either basal or prismatic slip. Edge dislocations, induced by misfit dislocations, favor the c-axis, remaining well confined within the in-plane interfacial layer of the m-plane α-Ga2O3, leading to reduced low edge dislocation density in the subsequent thick epitaxial layer. These findings shed light on the epitaxial dynamics of α-Ga2O3 heteroepitaxy, paving the way for the development of high-performance power devices.

Iodine migration in lead halide perovskite solar cells investigated by X-ray photoelectron spectroscopy

Abstract We investigate iodine migration in lead halide perovskite solar cells using X-ray photoelectron spectroscopy. We observe iodine migration to the surface direction at both the solar cell’s electrode and non-electrode regions. Our findings suggest that diffusion can drive iodine migration even without an electric field. We also detected iodine accumulation near the Ag electrode, probably related to increased series resistance and reduced efficiency. Present results highlight the need for a dense interfacial layer to suppress iodine migration.

Low power tactile sensory neuron using nanoparticle-based strain sensor and memristor

Endowing strain sensors with neuromorphic computing capabilities could permit the efficient processing of tactile information on the edge. The realization of such functionalities from a simple circuit without software processing holds promise for attaining skin-based perception. Here, leveraging the intrinsic neuronal plasticity of memristive neurons, various firing patterns induced by the applied strain were demonstrated. More specifically, tonic, bursting, transition from tonic to bursting, adaptive, and nociceptive activities were captured. The implementation of these patterns permits the facile translation of the analog pressure signals into digital spikes, attaining accurate perception of various tactile characteristics. The tactile sensory neuron consisting of an RC circuit was composed of a SiO2-based conductive bridge memristor exhibiting leaky integrate-and-fire properties and a Pt nanoparticles (NPs)-based strain sensor with a gauge factor of ∼270. A dense layer of Pt NPs was also used as the bottom electrode for the memristive element, yielding the manifestation of a threshold switching mode with a switching voltage of only ∼350 mV and an exceptional switching ratio of 107. Our work provides valuable insights for developing low power neurons with tactile feedback for prosthetics and robotics applications.

Density Modulated Vertical Carbon Nanotube Architectures with Bolometric Effect

Novel density‐modulated carbon nanotube (CNT) blocks with controlled and tunable CNT densities in adjacent layers have been developed. Regions with varying densities are laterally patterned into different shapes with submicron resolution, enhancing the fabrication flexibility of new 3D nanoelectromechanical systems for diverse sensing applications. This technology platform adjusts lateral electrical resistance, mechanical properties such as effective Young's modulus, and both lateral and vertical thermal conductivity, which can vary by several orders of magnitude. Highlighting its potential, the CNTs exhibit broadband blackbody absorption from ultraviolet (UV) to terahertz (THz). The initial bolometric detector demonstrates features such as a voltage responsivity = 20.5 V W−1, a response time of less than 0.1 ms, measured robust operation up to 200 °C, with fabricated device dimensions of 20 × 30 μm2, and a low‐cost design suitable for mass production. Further optimizations of the lateral design can reduce the device dimensions to as small as 5 × 5 μm2 and improve the absorption in the main resistance region. Thus, this architecture provides a platform technology to increase the responsivity of the fabricated new 3D‐based bolometer devices by several orders of magnitude. Tiny objects such as biological cells can be characterized in real time.

Development of gel-like, high encapsulation efficiency and antibacterial cinnamaldehyde-loaded Pickering emulsion stabilized by EGCG enhanced whey protein isolate-gum arabic ternary nanocomplex

Pickering emulsions (PEs) loaded with essential oils have proven to be a promising delivery system in food preservation, thus attracting increasing research attention. In this study, epigallocatechin gallate (EGCG) enhanced whey protein isolate(WPI)-gum Arabic(GA) ternary spherical nanocomplex (WGE) was fabricated by thermal and pH double-induced method and used to stabilize antibacterial, antioxidant and sustained release Pickering emulsions loaded with cinnamaldehyde. The WGE nanocomplex exhibited biphasic surface wettability (78.2±2.8°), 1.73 times that of WPI, as well as outstanding interfacial tension (5.60 mN/m), 44.6 % lower than that of WPI-GA, mainly due to the electrostatic interactions between WPI and GA and enhanced surface hydrophobicity causing by EGCG, indicating its superb ability to stabilize Pickering emulsions. Using WGE nanocomplex as the Pickering emulsion stabilizer, PEs template was constructed. Confocal laser scanning microscopy (CLSM) showed the formation of a dense oil-water interface layer and gel-like network structure, which demonstrated good storage stability against creaming and coalescence, benefiting to cinnamaldehyde encapsulation. Finally, cinnamaldehyde was encapsulated effectively using this PEs pattern with high encapsulation efficiency under different conditions. The cinnamaldehyde Pickering emulsions (CPEs) demonstrated superior antioxidant ability against DPPH (>85 %) and ABTS (>78 %), as well as effective antibacterial capability against E. coli (>99.9999 %) and S. aureus (>99.99 %) than pure cinnamaldehyde. Moreover, CPEs showed slow sustained-release ability, which could satisfyingly prolong the biological activity of cinnamaldehyde. Therefore, the WGE stabilized cinnamaldehyde Pickering emulsions fabricated in this work might provide a promising alternative for the delivery of antibacterial and controlled release essential oils in the food industry.

Employing Xception convolutional neural network through high-precision MRI analysis for brain tumor diagnosis

The classification of brain tumors from medical imaging is pivotal for accurate medical diagnosis but remains challenging due to the intricate morphologies of tumors and the precision required. Existing methodologies, including manual MRI evaluations and computer-assisted systems, primarily utilize conventional machine learning and pre-trained deep learning models. These systems often suffer from overfitting due to modest medical imaging datasets and exhibit limited generalizability on unseen data, alongside substantial computational demands that hinder real-time application. To enhance diagnostic accuracy and reliability, this research introduces an advanced model utilizing the Xception architecture, enriched with additional batch normalization and dropout layers to mitigate overfitting. This model is further refined by leveraging large-scale data through transfer learning and employing a customized dense layer setup tailored to effectively distinguish between meningioma, glioma, and pituitary tumor categories. This hybrid method not only capitalizes on the strengths of pre-trained network features but also adapts specific training to a targeted dataset, thereby improving the generalization capacity of the model across different imaging conditions. Demonstrating an important improvement in diagnostic performance, the proposed model achieves a classification accuracy of 98.039% on the test dataset, with precision and recall rates above 96% for all categories. These results underscore the possibility of the model as a reliable diagnostic tool in clinical settings, significantly surpassing existing diagnostic protocols for brain tumors.

Internal Electron-Donation Allocation Design for Intrinsic Solubilization of Lithium Nitrate in Ester Electrolyte for Stable Lithium Metal Batteries.

Lithium metal batteries (LMBs) have become a hot topic in the research of next-generation advanced battery technology due to their high specific energy. However, the high reaction activity between lithium metal and electrolyte is considered one of the key bottlenecks limiting large-scale applications of LMBs. As a classic electrolyte additive, LiNO3 significantly improves the stability of lithium metal in ether-based electrolytes. However, its solubility in carbonate-based electrolytes widely used in lithium-ion batteries is extremely low, and its protective effect on lithium metal is limited, which has become a key obstacle to the commercial application of lithium metal batteries. Here, we enhanced the local negative charge density of carbonyl oxygen atoms in carbonate molecules by introducing electron donors, making it easier for them to coordinate with Li+, thereby weakening the interaction between Li+ and NO3-, and significantly increasing the solubility of LiNO3 in ester electrolytes. The modified ester solvent promotes the derivatization and decomposition of salt anions, leading to the formation of a dense SEI layer rich in LiF and LiNxOy. This significantly improves the stability of lithium metal in ester-based electrolytes. The assembled Li||Li symmetric battery shows excellent cycling performance of over 4000 hours.

Exploring protein natural diversity in environmental microbiomes with DeepMetagenome.

Protein natural diversity offers a vast sequence space for protein engineering, and deep learning enables its detection from metagenomes/proteomes without prior assumptions. DeepMetagenome, a Python-based method, explores protein diversity through modules for training and analyzing sequence datasets. The deep learning model includes Embedding, Conv1D, LSTM, and Dense layers, with sequence feature analysis for data cleaning. Applied to metallothioneins from a database of over 146 million coding features, DeepMetagenome identified over 500 high-confidence metallothionein sequences, outperforming DIAMOND and CNN-based models. It showed stable performance compared to a Transformer-based model over 25 epochs. Among 23 synthesized sequences, 20 exhibited metal resistance. The tool also successfully explored the diversity of three additional protein families and is freely available on GitHub with detailed instructions.

A practical Approach to Handwritten Digit Generation with Generative Adversarial Networks on MNIST

Implementation of a Generative Adversarial Network (GAN) utilizing the MNIST (Modified National Institute of Standards and Technology database) dataset of handwritten digits, the GAN comprises a generator creating synthetic images from random noise and a discriminator to classifying these images as real or fake. The Generator utilizes leaky ReLU (rectified linear unit) activations, Batch normalization and reshaping to produce 28x28 grayscale images, while the Discriminator uses dense layers, leaky ReLU, and dropout to enhance classification accuracy. Both networks are trained using the Adam optimizer to improve stability and performance. The GAN is trained in an adversarial setup where the Generator seeks to create convincing images and the Discriminator aims to correctly classify them. Results demonstrate the GAN's ability to generate high-quality handwritten digit images, displaying its effectiveness and providing valuable insights into best practices for GAN implementation and training.

CAD-PsorNet: deep transfer learning for computer-assisted diagnosis of skin psoriasis

Psoriasis, being a chronic, inflammatory, lifelong skin disorder, has become a major threat to the human population. The precise and effective diagnosis of psoriasis continues to be difficult for clinicians due to its varied nature. In northern India, the prevalence of psoriasis among adult population ranges from 0.44 to 2.8%. Chronic plaque psoriasis accounts for over 90% of cases. This study utilized a dataset of 325 raw images collected from a reputable local hospital using a digital camera under uniform lighting conditions. These images were processed to generate 496 image patches (both diseased and normal), which were then normalized and resized for model training. An automated psoriasis image recognition framework was developed using four state-of-the-art deep transfer learning models: VGG16, VGG19, MobileNetV1, and ResNet-50. The convolutional layers adopted various edge, shape, and color filters to generate the feature map for psoriasis detection. Each pre-trained model was adapted with two dense layers, one dropout layer, and one output layer to classify input images. Among these models, MobileNetV1 achieved the best performance, with 94.84% sensitivity, 89.37% specificity, and 97.24% overall accuracy. Hyper-parameter tuning was performed using grid search to optimize learning rates, batch sizes, and dropout rates. The AdaGrad (Adaptive gradient)) optimizer was chosen for its adaptive learning rate capabilities, facilitating quicker convergence in model performance. Consequently, the methodology’s performance improved to 94.25% sensitivity, 96.42% specificity, and 99.13% overall accuracy. The model’s performance was also compared with non-machine learning-based diagnostic methods, yielding a Dice coefficient of 0.98. However, the model’s effectiveness is dependent upon high-quality input images, as poor image conditions may affect accuracy, and it may not generalize well across diverse demographics or psoriasis variations, highlighting the need for varied training datasets for robustness.