Deep Process Research Articles

Marketing’s crystal ball: Where we are and where we could soon be with generative artificial intelligence in marketing

Generative artificial intelligence (GenAI) has emerged as a transformative force in contemporary marketing, offering unprecedented creativity and efficacy. From predictive analytics to real-time personalisation, consumer behaviour patterning to personalised content creation, GenAI empowers marketers to craft bespoke strategies tailored to individual needs. As deep learning and natural language processing advance, GenAI promises to revolutionise brand–consumer relationships, forging authentic connections in the digital ecosystem. This paper studies the rise of GenAI in marketing, weaving research insights and visionary foresight. The study provides practitioners with an overview of GenAI’s potential for enhancing marketing efficacy and insight into its possible applications.

OPC UA-based Force Control for Deep Rolling with Mechanical Tools

Deep rolling can increase the service life of components under cyclic loads. This is achieved by plastic deformation of the component surface using a roller. The degree of plastic deformation depends on the rolling force. For a maximum increase in service life, the rolling force applied to the roller must stay within defined limits. When performing deep rolling using mechanical deep rolling tools, the rolling force is applied via an elastic element such as a leaf spring. The deep rolling force therefore depends on the deflection of leaf spring. It varies when the workpiece bends or has a varying diameter. These variations in the deep rolling force are not desired. They must be reduced to achieve optimum process results. To achieve this, a novel deep rolling force control for mechanical deep rolling was introduced in this paper. To control the rolling force, the force was measured with a sensory tool and machine tool axes were adjusted accordingly. The adjustment of the axes was implemented using the OPC UA standard. It is shown that this can reduce the force error during mechanical deep rolling by 90%. This ensures a constant process result and improves the mechanical deep rolling process.

The relationship between the paleobiogeography of the northern and southern sides of the Neotethys and the deep geodynamic processes

The relationship between the paleobiogeography of the northern and southern sides of the Neotethys and the deep geodynamic processes

Robust Forest Fire Detection using Deep Convolutional Neural Networks

Forest fires pose significant threats to ecosystems, wildlife, and human lives, necessitating proactive measures for early detection and rapid response. This paper presents FireGuard, an efficient model for forest fire detection using deep convolutional neural networks (CNNs). Leveraging the power of deep learning and image processing techniques, FireGuard analyzes aerial imagery and satellite data to detect signs of smoke and fire outbreaks in forested areas. The model utilizes a lightweight CNN architecture optimized for real-time performance and resource-constrained environments, making it suitable for deployment on unmanned aerial vehicles (UAVs), surveillance cameras, and satellite platforms. Experimental results demonstrate the effectiveness of FireGuard in accurately identifying forest fires with high precision and recall, outperforming traditional methods and existing deep learning models. By providing early warnings of potential fire incidents, FireGuard enables timely intervention by fire fighting agencies, thereby mitigating the impact of forest fires and preserving natural habitats.

Некоторые современные способы переработки медеплавильного шлака

An analysis of some modern methods of processing copper smelter slags containing iron, copper and other valuable components was carried out. Obviously, for deep complex processing, a combination of pyro and hydrometallurgical methods is necessary in order to use the positive qualities of both methods. So iron in the form of a finished product of iron and steel can only be extracted by pyrometallurgical methods, and non-ferrous metals are extracted using hydrometallurgical methods. The use of force fields at the leaching stage can also have a positive effect on the efficiency of processing.

Analog Implementation of a Spiking Neuron with Memristive Synapses for Deep Learning Processing

Analog neuromorphic prototyping is essential for designing and testing spiking neuron models that use memristive devices as synapses. These prototypes can have various circuit configurations, implying different response behaviors that custom silicon designs lack. The prototype’s behavior results can be optimized for a specific foundry node, which can be used to produce a customized on-chip parallel deep neural network. Spiking neurons mimic how the biological neurons in the brain communicate through electrical potentials. Doing so enables more powerful and efficient functionality than traditional artificial neural networks that run on von Neumann computers or graphic processing unit-based platforms. Therefore, on-chip parallel deep neural network technology can accelerate deep learning processing, aiming to exploit the brain’s unique features of asynchronous and event-driven processing by leveraging the neuromorphic hardware’s inherent parallelism and analog computation capabilities. This paper presents the design and implementation of a leaky integrate-and-fire (LIF) neuron prototype implemented with commercially available components on a PCB board. The simulations conducted in LTSpice agree well with the electrical test measurements. The results demonstrate that this design can be used to interconnect many boards to build layers of physical spiking neurons, with spike-timing-dependent plasticity as the primary learning algorithm, contributing to the realization of experiments in the early stage of adopting analog neuromorphic computing.

Interaction between shallow and deep structures: Forward-modeling kinematics sequence of a fold-and-thrust belt with double detachment

It is crucial to determine the sequence of deformation and the interaction between shallow and deep structures within the multiple detachment systems to comprehend geological processes fully in fold-and-thrust belts (FTBs). Here, we implement forward modeling to investigate the geometric, kinematic, and syntectonic sedimentary characteristics of shallow and deep structural interactions. We design four combinative patterns of geometric structures, three kinematic sequence scenarios, and two settings with different ratios of the sedimentation rate to the uplift rate to examine the influence of such settings on growth strata geometry. The most effective results of 24 groups of shallow and deep structural interactions were obtained for various combinations of fault-bend fold, detachment fold, and tectonic wedge. This study provides a sequential kinematic image of FTBs, illustrating the interaction of two sets of detachments and demonstrating how different styles of pre- and growth strata can be utilized to identify shallow and deep structure deformation processes. The geometry of pre- and growth strata observed in the models is similar to that observed in natural structures. These forward models facilitate a better understanding of the complex geometry and kinematics of the interaction between shallow and deep fault-related folds.

Design of Multidomain Feature Analysis Model for Estimation of Anemic Conditions via Deep Transfer Learning

The detection of anemia from anemic retinopathy involves the identification and analysis of specific retinal changes that occur as a result of severe anemia which requires analysis of multimodal sources. Existing models for identification of these conditions either showcase high complexity, or have lower efficiency when evaluated on clinical scans. Moreover, most of these models cannot be scaled for multi-disease images, which limits their applicability under real-time use cases. To overcome these issues, this research article proposes the design of an efficient and novel multidomain feature analysis model for the estimation of anemic conditions with the help of deep transfer learning process. The proposed model initially represents retinal image scans into multiple domains via estimation of Frequency, Gabor, Convolution, Wavelet and Entropy features. These features are selected via Bacterial Foraging Optimizer (BFO), which assists in retaining high variance feature sets. Due to this, the model is able to reduce redundancies, that assists in improving the speed and accuracy of classification operations. To perform this task, the model uses a set of customized 1-D Cascaded Binary CNN, that assist in categorizing input images into 4 different classes. These include, “Anemic due to Glaucoma”, “Glaucoma”, “Anemic”, and “Normal”, with 98.3% accuracy which is 4.5% higher than existing methods. The model is also able to improve the precision by 2.9%, recall by 3.5%, while reducing the delay needed for classification by 5.9% when compared with existing techniques. Due to which, the model is highly useful for clinical deployments.

Quantification of litter in cities using a smartphone application and citizen science in conjunction with deep learning-based image processing

Cities are a major source of litter pollution. Determination of the abundance and composition of plastic litter in cities is imperative for effective pollution management, environmental protection, and sustainable urban development. Therefore, here, a multidisciplinary approach to quantify and classify the abundance of litter in urban environments is proposed. In the present study, litter data collection was integrated via the Pirika smartphone application and conducted image analysis based on deep learning. Pirika was launched in May 2018 and, to date, has collected approximately one million images. Visual classification revealed that the most common types of litter were cans, plastic bags, plastic bottles, cigarette butts, cigarette boxes, and sanitary masks, in that order. The top six categories accounted for approximately 80 % of the total, whereas the top three categories accounted for more than 60 % of the total imaged litter. A deep-learning image processing algorithm was developed to automatically identify the top six litter categories. Both precision and recall derived from the model were higher than 75 %, enabling proper litter categorization. The quantity of litter derived from automated image processing was also plotted on a map using location data acquired concurrently with the images by the smartphone application. Conclusively, this study demonstrates that citizen science supported by smartphone applications and deep learning-based image processing can enable the visualization, quantification, and characterization of street litter in cities.

Federated Learning with Homomorphic Encryption for Ensuring Privacy in Medical Data

Federated Learning (FL) is a machine learning methodology that allows remote devices to collectively train a learning system without sharing their data. FL-based methods provide enhanced secure privacy by transmitting only localized model variables, learned with local information, from dispersed devices to a centralized controller. There is a potential for a centralized server or malicious individuals to deduce or get sensitive private data by analyzing the structure and variables of regional learning networks. This study incorporates the FL process into the deep learning process of medical prototypes in an Internet of Things (IoT)-based medical facility. A Secured Medical Homomorphic Encryption Algorithm (SMHEA) is proposed in this research to ensure medical data privacy. Cryptographic primitives, such as masking and homomorphic cryptography, are used to enhance the security of local modeling. This prevents adversaries from deducing confidential health information via assaults like model restoration or modeling inversion. The primary determinant for assessing the regional modeling's contributions to the universal model during each training stage is the quality of the databases possessed by various individuals rather than the typically used metric of database dimension in deep learning. A dropout-tolerant approach is suggested, where the FL procedure would continue as long as the total amount of online customers remains over a certain level. By doing a security evaluation, it is evident that the suggested approach effectively ensures data privacy. Theoretical analysis is conducted on accuracy, computation time, and communication error. An example of clinical applications is the categorization of skin lesions using training photos from the HAM10000 medical database. The experimental findings demonstrate that the suggested system had favorable performance and privacy preservation outcomes compared to current methods.

Leveraging Large Language Models for Knowledge-free Weak Supervision in Clinical Natural Language Processing.

The performance of deep learning-based natural language processing systems is based on large amounts of labeled training data which, in the clinical domain, are not easily available or affordable. Weak supervision and in-context learning offer partial solutions to this issue, particularly using large language models (LLMs), but their performance still trails traditional supervised methods with moderate amounts of gold-standard data. In particular, inferencing with LLMs is computationally heavy. We propose an approach leveraging fine-tuning LLMs and weak supervision with virtually no domain knowledge that still achieves consistently dominant performance. Using a prompt-based approach, the LLM is used to generate weakly-labeled data for training a downstream BERT model. The weakly supervised model is then further fine-tuned on small amounts of gold standard data. We evaluate this approach using Llama2 on three different n2c2 datasets. With no more than 10 gold standard notes, our final BERT models weakly supervised by fine-tuned Llama2-13B consistently outperformed out-of-the-box PubMedBERT by 4.7-47.9% in F1 scores. With only 50 gold standard notes, our models achieved close performance to fully fine-tuned systems.

Development of the formulation and technology of a new fish culinary product "Cream cheese fish from blue catfish" enriched with iodine of kelp of the White Sea

Expanding the range of products based on the rational use of available fish resources is a current direction for the development of the food industry of the Russian Federation. Blue catfish, a little-used fishery item in the Northern Basin, was chosen as a raw material for the development of a fish culinary paste product. Blue catfish meat is significantly watered (up to 93 % of the total mass), as a result there are currently no technologies for deep processing of fish for food purposes. During the research, the technology of a new fish culinary product "Fish cream cheese from blue catfish, enriched with iodine from White Sea kelp" has been proposed in an assortment, and recipes have been developed, optimized in the MatLab program. The amino acid composition of blue catfish meat as well as indicators of the nutritional value of new products have been considered. The energy value of 100 g of product varies from 120 to 150 kcal, the mass fraction of protein is 8.5–11 %; fat – 9–13 %; carbohydrates – 2–2.5 %. The iodine content in a unit of consumer packaging of the product "Cream cheese fish from blue catfish with pickled seaweed" was experimentally established to be 340 mcg per 100 g; a 44 g serving of this product contains 100 % of the recommended daily intake of the micronutrient (150 mcg). The developed product "Cream cheese fish from blue catfish with pickled seaweed" can be classified as "enriched with iodine" when the amount of its daily consumption (25 g) is indicated on the consumer packaging.

Toward a better understanding of China's e-waste reverse logistics implementation barriers: A certified disassembler's perspective

E-waste, containing valuable materials and hazardous substances, poses a significant challenge to the environment and society. One approach to addressing this challenge is the implementation of a reverse logistics network to recover e-waste, which involves a disassembler connecting the upstream collector with the downstream recycler and disposer. In China, a certified disassembler in a subsidy scheme plays a crucial role in e-waste recycling effectiveness. However, a declining recycling rate and high idle capacity have led to a ‘certified disassembler dilemma’. This study uses stakeholder analysis processes to identify barriers to e-waste reverse logistics, and the grey-based decision-making trial and evaluation laboratory approach is utilized to explore their causal relationships. The results, supported by a sensitivity analysis, indicate that the certified disassemblers' internal capability deficiencies form the most significant causal barrier, with a net cause value of 2.74. In contrast, the increased number of illegal disassemblers is the primary determinant of all other barriers, with a prominence value of 19.5. Recommendations for improving certified disassemblers' internal capabilities are provided, including exploring the deep processing of waste materials, extending the residential collection network, and reassessing multiple e-waste recovery options.

Novel core shell structure to preclude phase segregation of iron-based oxygen carriers for chemical looping combustion

Iron-based oxygen carriers as one of the most commonly used oxygen carriers in chemical looping combustion, shows poor cycling stability. In this work, spinel-structured, Fe2O3@MgO core–shell structure, and Fe2O3/CuO@MgO core–shell structure oxygen carriers were prepared using compression molding. The effects of deep reduction and oxidative regeneration processes on the cyclic stability of oxygen carriers with different structures were investigated using thermal gravimetric analyzer. The mechanisms for the changes in the cyclic stability were explored through a series of characterization analyses. The results showed that the reduction process of spinel oxygen carriers generated Fe-Mg solid solution, which was unfavorable to deep reduction, and phase separation occurred during 5 cycles. Due to the migration of Fe3+ ions to the surface, surface was seriously sintered, deteriorating the cyclic stability of spinel oxygen carriers. For Fe2O3@MgO oxygen carrier, the deep reduction can be realized, but the interdiffusion phenomenon of Fe2+ and Mg2+ occurred after 5 reductions, and the generation of (MgO)x(FeO)1−x solid solution impeded the deep reduction. The addition of CuO in the core of oxygen carriers to form Fe2O3/CuO@MgO could inhibit the interdiffusion of Fe2+ and Mg2+, and maintained the oxygen carrying capacity up to more than 95 % in 10 cycles. Meanwhile, CuO synergized with Fe2O3 to enhance the reduction and oxidation reactivity of the oxygen carrier. On this basis, the ion diffusion mechanism during the deep reduction and oxidation of iron-based oxygen carriers with different structures was proposed. This study provides guidance for the modification of iron-based oxygen carriers with superior cycling stability.

Advances in the Preparation, Structure and Bioactivity of Polysaccharides from Lycium ruthenicum Murr.: A Review.

Lycium ruthenicum Murr. is rich in polysaccharides, and the polysaccharides in Lycium ruthenicum Murr. (LRPS) have various bioactivities, such as antioxidant activity, anti-tumor activity, neuroprotective activity, and immunomodulatory activity. It has broad prospects in the development of functional foods and pharmaceuticals. Researchers have found that the structural characteristics of LRPS, such as molecular weight, monosaccharide composition, primary structure, etc., have a significant impact on their bioactivities. Therefore, studying the structure of LRPS is of great significance in revealing their bioactivities and mechanisms. This study, based on introducing the preparation methods of LRPS, focuses on reviewing the research progress on the main structural characteristics, various bioactivities, and mechanisms of action of LRPS. In addition, the study provides prospects for the development of LRPS in the fields of food and medicine, aiming to provide theoretical support for its deep processing and application.

Enzyme and Microwave Coassisted Extraction, Physicochemical Properties, and Antioxidant Activity of Polysaccharides from Baijiu Distillers' Grains

AbstractBaijiu distillers' grain is a solid waste after the fermentation of grains, and their high added value and deep processing and utilization require further study. This study uses the enzyme and microwave coassisted method to extract Baijiu distillers' grain polysaccharides (DGPs). Through optimization of the single‐factor test and response surface methodology (RSM), the optimal conditions are solid–liquid ratio 1:30, enzyme time 50 min, enzyme temperature 50 °C, enzyme pH 5.0, microwave time 6 min, and microwave power 480 W, with a maximum DGP yield of 3.98% ± 0.03. DGP is purified by DEAE‐Sepharose cellulose ion exchange chromatography to obtain four fractions, namely DGP‐1, DGP‐2, DGP‐3, and DGP‐4; the molecular weights are 2224.05, 90.99, 1478.16, and 346.19 kDa, respectively, and the main monosaccharide components are Man and Glc. The antioxidant assay shows that the four kinds of DGPs have different free radical scavenging capacities, among which DGP‐2, DGP‐3, and DGP‐4 have good antioxidant activities. Therefore, DGP can be used as a potential natural antioxidant.

Modeling of the multi-cycle deep charge and discharge attenuation mechanism of non-aqueous Li–O2 batteries

Due to the high energy density, non-aqueous lithium-oxygen (Li–O2) batteries attract significant attention. However, batteries' high capacity attenuation rate under deep charge and discharge conditions remains a significant challenge. This paper presents a multi-cycle deep charge and discharge model for non-aqueous lithium-oxygen batteries, which predicts the performance of the battery during multiple deep charge and discharge cycles at high and low discharge-specific capacities. The parameter states during different discharge stages in different discharge cycles are investigated by analyzing the battery's cathode porosity, product volume fraction, and oxygen concentration changes. The study shows that as the number of cycles increases, the deposition of a small number of discharge products in the cathode altered the distribution of the newly generated products, thereby affecting the cathode structure and oxygen transport in the subsequent discharge. Moreover, depositing a small amount of discharge products can result in significant capacity attenuation in the battery. This model can accurately evaluate the deep charge and discharge performance attenuation process of non-aqueous Li–O2 batteries, which helps improve the understanding of the deep discharge attenuation mechanism of non-aqueous Li–O2 batteries.

The hidden link between dry ankaramitic melts and giant alkalic-type epithermal Au deposits: The Lihir Island example

The magmatic controls on the formation of giant epithermal Au deposits genetically associated with alkaline magmatic systems are not well understood. The general model, which attributes the generation of alkaline magmas to low-degree partial melting of metasomatized lithospheric mantle in a post-subduction tectonic setting cannot easily explain why some of these magmas are highly fertile for Au mineralization while others are not. This knowledge gap hampers the effective exploration of this economically important deposit type. In this study, we focused on the alkaline magmatism on Lihir Island and the Conical Seamount, Papua New Guinea, that gave birth to the Ladolam deposit and investigated the composition of silicate melt inclusions (SMIs) hosted in primitive volcanic rocks. Our findings offer important and novel insights into the early and deep magmatic processes that controlled the ore fertility of the melts that produced the largest known alkalic-type epithermal Au deposit on Earth. Olivine- and clinopyroxene-hosted primitive SMIs reveal nepheline-normative ankaramitic melt compositions derived from high-degree partial melting of metasomatized clinopyroxene-dominated lithologies, presumably amphibole-bearing clinopyroxenitic cumulates and metasomatically overprinted lithospheric mantle, both produced during earlier active subduction. These ultra-calcic parental melts are relatively low in H2O, S, and Cl, but enriched in Cu and highly oxidized. Due to their low water contents, significant clinopyroxene crystallization over a narrow temperature interval drove melt evolution from ankaramitic to alkaline compositions by strongly increasing incompatible element concentrations. Despite initially low H2O concentrations, evidence for high-temperature degassing is present in the form of low-density single-phase fluid inclusions co-existing with SMIs in clinopyroxene. This degassing event led to a strong decrease of S concentrations in the melt, whereas Cl and chloride-complexing elements remained largely unaffected. We propose that high-temperature volatile saturation as consequence of significant clinopyroxene crystallization is a key process for subsequent epithermal Au mineralization and can explain the efficient separation of Au from Cu. The complex tectonic setting of Lihir Island, marked by a reversal of subduction polarity, a tear in the slab beneath the island, and trans-lithospheric extensional structures, facilitated the generation and ascent of highly oxidized and relatively dry ankaramitic melts that ultimately exsolved a vapor-like, S- and presumably Au-rich fluid phase and might thus represent a prerequisite for alkalic-type epithermal Au mineralization.

Separation of quinoline from quinoline/2-methylnaphthalene using deep eutectic solvents: Experimental and COSMO-RS prediction

The high-temperature coal tar wash oil contains two valuable components, namely, quinoline and 2-methylnaphthalene. These two components are challenging to separate during the deep processing. To address this issue, the COSMO-RS (Conductor-like Screening Mode for Real Solvent) model was utilized to identify them as a potential deep eutectic solvents (DESs) extractant. Based on the calculated infinite dilution activity coefficient, the performance indexes of DESs, including selectivity, solubility, and performance index were determined and utilized to identify high-performance extractants for the quinoline/2-methylnaphthalene system. Finally five DESs were selected based on these indexes, they are DES1 formed by Choline chloride: propionic acid with the molar ratio of 1:1, similarly, DES2 is Choline chloride:oxalic acid of 1:1, DES3 Tetraethylammonium chloride:propionic acid of 1:1, DES4 Tetraethylammonium bromide:itaconic acid of 1:2 and DES5 Tetraethylammonium bromide:propionic acid of 1:2. The liquid–liquid phase equilibrium experiments of the aforementioned five DESs at different temperatures (30 ℃, 40 ℃, and 50 ℃) under atmospheric pressure were conducted. The results showed that five kinds of DESs had good extraction effect on quinoline. The DESs formed by choline chloride and anhydrous oxalic acid had the best extraction effect. The quantum chemical single-molecule optimization of different HBDs and HBAs in DESs, intermolecular interaction energy analysis and independent gradient modeling shows that it is mainly induced by the formation of shared electron pairs by providing empty orbitals from DESs and lone pair of electrons from the quinoline molecule, and the main force is the weak hydrogen bonding force.

Stored energy density solution for TSV-Cu structure deformation under thermal cyclic loading based on PINN

TSV-Cu is widely used for chip interconnects, where high testing costs and complex crystal plasticity finite element (CPFE) limit the research of its deep microplastic evolution process. Considering the stored energy density (SED) based on dislocation density and plastic work as a fatigue indicator factor (FIP), this paper proposes for the first time a physics-informed neural network (PINN) framework based on SED, which aims to achieve the solution of SED distributions quickly and efficiently to save computational time and cost. The grain orientation, geometrical compatibility factor, back stress, and effective plastic strain are taken into account as inputs to the PINN model. The total dislocation density is used as an indirect solution variable to construct the associated loss boundary terms, which results in the solution of the SED. The results of the two real EBSD tests show that the PINN model is able to accurately and sensitively predict the SED concentration distribution for different thermal cycle loadings, and maintains a high degree of agreement with the CPFE calculation results. Moreover, The superiority of PINN over other machine learning algorithms in terms of physical model interpretation and prediction accuracy is verified. These make it a reality for PINN to solve for the FIP distribution for the first time, and to accurately and quickly predict the location of crack initiation.