Distillation Method Research Articles

Graphene oxide aerogels for adsorptive separation of aromatic hydrocarbons and cycloalkanes

Efficient separation of benzene and cyclohexane has critical importance for production of commodity chemicals, and is one of the most challenging separations in the industry. Physisorption by recyclable, porous solids has a significant potential in substituting energy-intensive azeotropic or extractive distillation methods. Reduced graphene oxide aerogels (rGOAs) are emerging materials holding great promise for connecting unique properties of 2D graphene with ordinary 3D materials. The benzene/cyclohexane separation on rGOAs self-assembled by the chemical reduction with l-ascorbic acid, sodium bisulphite and (for the first time) sodium dithionite was studied by dynamic gas adsorption methods, and the adsorption performance was analysed in relation to aerogels physicochemical properties. The aerogel reduced with sodium dithionite (rGOA_DTN) had the highest reduction degree and specific surface area (461.2 m2g-1), with the highest contribution of mesopores. It was also the sample with the uppermost uptake of benzene and cyclohexane. The binary component adsorption on rGOA_DTN resulted in the selectivity of the adsorption of benzene over cyclohexane of 2.1. Adsorption-desorption studies demonstrated the excellent thermal stability of the adsorbent in the long-run operation. Because the adsorption capacity did not correlate with the mesopores but with macropores surface area, the selectivity of the adsorption was attributed to the different physicochemical structure of aerogels surface. The benzene molecule interacted strongly by specific C-H···π interactions, while the cyclohexane molecule was excluded from the surface of aerogels because of its shape/size. Results demonstrated that rGOAs can be a versatile and flexible platform for adsorptive gas-phase hydrocarbons separation.

Learning From Human Educational Wisdom: A Student-Centered Knowledge Distillation Method.

Existing studies on knowledge distillation typically focus on teacher-centered methods, in which the teacher network is trained according to its own standards before transferring the learned knowledge to a student one. However, due to differences in network structure between the teacher and the student, the knowledge learned by the former may not be desired by the latter. Inspired by human educational wisdom, this paper proposes a Student-Centered Distillation (SCD) method that enables the teacher network to adjust its knowledge transfer according to the student network's needs. We implemented SCD based on various human educational wisdom, e.g., the teacher network identified and learned the knowledge desired by the student network on the validation set, and then transferred it to the latter through the training set. To address the problems of current deficiency knowledge, hard sample learning and knowledge forgetting faced by a student network in the learning process, we introduce and improve Proportional-Integral-Derivative (PID) algorithms from automation fields to make them effective in identifying the current knowledge required by the student network. Furthermore, we propose a curriculum learning-based fuzzy strategy and apply it to the proposed PID control algorithm, such that the student network in SCD can actively pay attention to the learning of challenging samples after with certain knowledge. The overall performance of SCD is verified in multiple tasks by comparing it with state-of-the-art ones. Experimental results show that our student-centered distillation method outperforms existing teacher-centered ones.

A New Framework of Collaborative Learning for Adaptive Metric Distillation.

This article presents a new adaptive metric distillation approach that can significantly improve the student networks' backbone features, along with better classification results. Previous knowledge distillation (KD) methods usually focus on transferring the knowledge across the classifier logits or feature structure, ignoring the excessive sample relations in the feature space. We demonstrated that such a design greatly limits performance, especially for the retrieval task. The proposed collaborative adaptive metric distillation (CAMD) has three main advantages: 1) the optimization focuses on optimizing the relationship between key pairs by introducing the hard mining strategy into the distillation framework; 2) it provides an adaptive metric distillation that can explicitly optimize the student feature embeddings by applying the relation in the teacher embeddings as supervision; and 3) it employs a collaborative scheme for effective knowledge aggregation. Extensive experiments demonstrated that our approach sets a new state-of-the-art in both the classification and retrieval tasks, outperforming other cutting-edge distillers under various settings.

Medical image classification: Knowledge transfer via residual U-Net and vision transformer-based teacher-student model with knowledge distillation

With the widespread integration of deep learning techniques in the domain of medical image analysis, there is a prevailing consensus regarding their efficacy in handling high-dimensional and intricate medical image data. However, it is imperative to acknowledge that while complex deep models exhibit a remarkable capacity for processing high-dimensional and intricate data, they often necessitate a substantial allocation of computational resources and time. Furthermore, lightweight models, despite their computational efficiency, tend to underperform when compared to their more intricate counterparts in terms of performance. Hence, the prevailing aspiration is to transfer the cognitive prowess of complex models to their lightweight counterparts. Addressing the aforementioned concern, this study proposes a knowledge distillation approach that encompasses joint feature and soft label transfer. It entails the transference of knowledge from the teacher model’s intermediate features and predictive outcomes to the student model. The student model leverages this knowledge to emulate the behavior of the teacher model, thereby enhancing the precision of its own predictions. Building upon this foundation, we introduce a Res-Transformer teacher model based on the U-Net architecture and a student model known as ResU-Net, which is grounded in residual modules. The Res-Transformer teacher model employs multi-layer residual attention during the downsampling process to capture deep-level features of the image. Subsequently, we have incorporated a Multi-layer Perceptual Attention module (MPA) for each skip connection layer, facilitating the integration of hierarchical upsampled information to restore fine-grained details within the feature maps. The ResU-Net student model enhances network stability through the utilization of residual modules and optimizes skip connections to recover any lost image information during convolutional operations. Lastly, we conducted experimental assessments on multiple disease datasets. The results reveal that the ACC of the Res-Transformer model achieves an impressive 96.9%. Furthermore, through the knowledge distillation method, rich knowledge is effectively transferred to the ResU-Net model, resulting in a remarkable ACC improvement of 7.2%.

The effect of pressure and reflux ratio on the spinning band distillation process on the purity of myristicin isolate from the terpenless fraction of nutmeg oil

Abstract Nutmeg oil contains myristicin, which functions as an antioxidant and antiaging in cosmetic products by inhibiting the activity of the collagenase enzyme. Isolation of myristicin compounds from the terpeneless fraction is a strategy to increase the purity of myristicin isolates effectively and efficiently. This research aims to identify the purity of the myristicin compound from the terpenless fraction of nutmeg oil and the yield based on different pressure and reflux ratio variables. Fractionation was carried out using the spinning band distillation method at pressures of 10 and 15 mmHg, with reflux ratios of 3:1 and 5:1. The fraction cutting temperature is regulated based on the boiling point of the compound starting from an open cut temperature of 172 °C to a close cut temperature of 300 °C. Parameters analyzed: The physicochemical properties of the material are specific gravity, refractive index, solubility in 96% ethanol, myristicin content, and yield. Research shows that the pressure and reflux ratio are not significant to the purity of myristicin, caused by the range of pressure and reflux ratio values being too close, but can increase the purity of myristicin compounds from initial levels of 51.25% to 83.23%. Thus, spinning band fractional distillation is able to be used as a tool for purifying the myristicin compound in the terpenless fraction of nutmeg oil.

Patchouli essential oil: Unlocking antimicrobial potential for dandruff control

Abstract Dandruff causes the scalp to peel and results in itching, which is caused by one of the fungi, Malassezia globosa. Several studies have revealed that the compounds in essential oils can act as antimicrobials. To broaden insights regarding the potential compounds in essential oils, this research focuses on the potential of patchouli essential oil in inhibiting the growth of dandruff-causing microbes. One of the compounds with antimicrobial potential in patchouli oil is patchouli alcohol. The steam distillation method was used for extraction oil from dried patchouli stems and leaves to produce three different types of oil samples: crude patchouli oil, heavy fraction patchouli oil, and light fraction patchouli oil. Subsequently, the GC-MS analysis was conducted to compare the compound contents in samples. Our results indicate that the three patchouli oil samples have varying patchouli alcohol contents. Crude patchouli oil contains 31.2% patchouli alcohol; heavy fraction patchouli oil contains 61.82% patchouli alcohol, while light fraction patchouli oil contains 23.99% patchouli alcohol. This study demonstrates that heavy fraction patchouli oil has a greater potential as an antimicrobial compared to crude patchouli oil and light fraction patchouli oil.

Adaptive Knowledge Distillation for High-Quality Unsupervised MRI Reconstruction With Model-Driven Priors.

Magnetic Resonance Imaging (MRI) reconstruction has made significant progress with the introduction of Deep Learning (DL) technology combined with Compressed Sensing (CS). However, most existing methods require large fully sampled training datasets to supervise the training process, which may be unavailable in many applications. Current unsupervised models also show limitations in performance or speed and may face unaligned distributions during testing. This paper proposes an unsupervised method to train competitive reconstruction models that can generate high-quality samples in an end-to-end style. Firstly teacher models are trained by filling the re-undersampled images and compared with the undersampled images in a self-supervised manner. The teacher models are then distilled to train another cascade model that can leverage the entire undersampled k-space during its training and testing. Additionally, we propose an adaptive distillation method to re-weight the samples based on the variance of teachers, which represents the confidence of the reconstruction results, to improve the quality of distillation. Experimental results on multiple datasets demonstrate that our method significantly accelerates the inference process while preserving or even improving the performance compared to the teacher model. In our tests, the distilled models show 5%-10% improvements in PSNR and SSIM compared with no distillation and are 10 times faster than the teacher.

Knowledge distillation of face recognition via attention cosine similarity review

AbstractDeep learning‐based face recognition models have demonstrated remarkable performance in benchmark tests, and knowledge distillation technology has been frequently accustomed to obtain high‐precision real‐time face recognition models specifically designed for mobile and embedded devices. However, in recent years, the knowledge distillation methods for face recognition, which mainly focus on feature or logit knowledge distillation techniques, neglect the attention mechanism that play an important role in the domain of neural networks. An innovation cross‐stage connection review path of the attention cosine similarity knowledge distillation method that unites the attention mechanism with review knowledge distillation method is proposed. This method transfers the attention map obtained from the teacher network to the student through a cross‐stage connection path. The efficacy and excellence of the proposed algorithm are demonstrated in popular benchmark tests.

Extraction of geraniol from palmarosa oil using hydrotropic solvents

The extraction of geraniol from palmarosa oil using hydrotropic solvents was investigated. Palmarosa oil possesses an appealing rose aroma and properties like anti-inflammatory, antifungal, and antioxidant due to the presence of geraniol. The extraction of geraniol from palmarosa oil by using distillation methods like steam distillation and fractional distillation was a laborious process. So hydrotropes were tried for extraction. The geraniol yield and purity depend on parameters like concentration of hydrotrope, solvent volume ratio, and time period. Using the Box Benkhem Design (BBD), the extraction process was optimized. One of the major advantages of using hydrotropic solvents is that they were classified as green solvents, and recovery of solvents is also possible. To reduce the extraction time probe sonication is carried out. Different hydrotropic solvents with probe sonication are done on palmarosa oil by altering various process parameters to study the separation, yield, and purity.

MV-MR: Multi-Views and Multi-Representations for Self-Supervised Learning and Knowledge Distillation.

We present a new method of self-supervised learning and knowledge distillation based on multi-views and multi-representations (MV-MR). MV-MR is based on the maximization of dependence between learnable embeddings from augmented and non-augmented views, jointly with the maximization of dependence between learnable embeddings from the augmented view and multiple non-learnable representations from the non-augmented view. We show that the proposed method can be used for efficient self-supervised classification and model-agnostic knowledge distillation. Unlike other self-supervised techniques, our approach does not use any contrastive learning, clustering, or stop gradients. MV-MR is a generic framework allowing the incorporation of constraints on the learnable embeddings via the usage of image multi-representations as regularizers. The proposed method is used for knowledge distillation. MV-MR provides state-of-the-art self-supervised performance on the STL10 and CIFAR20 datasets in a linear evaluation setup. We show that a low-complexity ResNet50 model pretrained using proposed knowledge distillation based on the CLIP ViT model achieves state-of-the-art performance on STL10 and CIFAR100 datasets.

Main Components of The Semboro Variety of Siam Orange Peel Essential Oil Prepared Using Water Distillation Method

The maximum production potential for low quality oranges is 15% or 9,057 tons/year, the potential for orange peel waste is 50% of the total weight of oranges or the equivalent of 4,500 tons of orange peel waste The average yield of orange peel essential oil is 0.6%, resulting in a potential of 27 tons/year. The aim of this research is to obtain information regarding the best heating temperature in the distillation process and to identify the main components of Siam orange peel essential oil of the Semboro variety. The research was carried out using a laboratory-scale water distillation system, with a distillation flask capacity of 1 liter, with 150 g of orange peel as raw material at a distillation process heating temperature of 250°C and 300°C. The yields of 1.9 mL and 0.9 mL of Siam orange peel essential oil were obtained 1.083% and 0.513% respectively. The specific gravity of Siam orange peel essential oil is 0.855 g/cm3 and 0.886 g/cm3 respectively, the main components are 29 and 18 components respectively. An important finding of this research is that at a distillation process with heating temperature of 250°C, yields a limonene content of 16.10%, whereas heating at 300°C results in 7.69% limonene content. Keywords: Limonene; Low grade orange peel; Heating temperature; Quality essential oil.

The effect of different hydrodistillation times on the composition and yield of bergamot (Citrus bergamia Risso) peel essential oil and a comparison of the cold‐pressing method

AbstractBergamot (Citrus bergamia Risso) is not consumed as fresh fruit as other members of Citrus. Bergamot essential oil has a unique aroma. Bergamot essential oil is a very valuable product used in perfumery, cosmetics, aromatherapy and food industry due to its beautiful scent and functional properties. As with other Citrus species, the composition of bergamot essential oil can vary according to harvest time, climate and extraction technology. Bergamot essential oil could be obtained by cold press extraction or distillation method. Within the scope of the study, the effect of four different hydrodistillation (HD) times (5, 10, 60 and 120 min) on the composition of bergamot essential oil obtained from Turkish fruits was investigated. In the first 60 min, 96.11% of the total essential oil was obtained. Research findings have shown that the distillation time is quite effective in the composition of bergamot essential oil. Limonene, linalool and linalyl acetate, which are the main components of bergamot essential oil, varied in the range, 42.36%–53.25%, 5.59%–12.18% and 20.84%–26.73%, respectively, with respect to distillation time. In cold press method, bergamot essential oil has lower linalool/linalyl acetate ratio (0.15) compared to HD method and a higher quality essence was obtained.

FedKG: A Knowledge Distillation-Based Federated Graph Method for Social Bot Detection.

Malicious social bots pose a serious threat to social network security by spreading false information and guiding bad opinions in social networks. The singularity and scarcity of single organization data and the high cost of labeling social bots have given rise to the construction of federated models that combine federated learning with social bot detection. In this paper, we first combine the federated learning framework with the Relational Graph Convolutional Neural Network (RGCN) model to achieve federated social bot detection. A class-level cross entropy loss function is applied in the local model training to mitigate the effects of the class imbalance problem in local data. To address the data heterogeneity issue from multiple participants, we optimize the classical federated learning algorithm by applying knowledge distillation methods. Specifically, we adjust the client-side and server-side models separately: training a global generator to generate pseudo-samples based on the local data distribution knowledge to correct the optimization direction of client-side classification models, and integrating client-side classification models' knowledge on the server side to guide the training of the global classification model. We conduct extensive experiments on widely used datasets, and the results demonstrate the effectiveness of our approach in social bot detection in heterogeneous data scenarios. Compared to baseline methods, our approach achieves a nearly 3-10% improvement in detection accuracy when the data heterogeneity is larger. Additionally, our method achieves the specified accuracy with minimal communication rounds.

Sustainable and green removal of arsenic from crude tin via vacuum gasification–directional condensation

In the realm of high–tech industries, Sn has emerged as an indispensable strategic metal. However, the pyrometallurgical refining process of crude Sn is marred by the production of a substantial amount of hazardous solid waste, notably As–Al slag, posing a direct threat to the ecological safety of the production areas. This study delineates a novel process for the removal of As from crude Sn, employing vacuum gasification–directional condensation, corroborated by theoretical calculations and laboratory experimental research. Theoretically, at a system pressure ranging from 1–10 Pa, the initial volatilization temperatures of Sn and As exhibit significant variance. However, the formation of intermetallic compounds can induce Sn volatilization. Experimentally, the condensation temperature for As was predominantly within 331.7–484.7 K. During the vacuum gasification–directional condensation process, the presence of Sn4As3 intermetallic compounds impeded As volatilization. And under optimal conditions, the As content in the Sn product was recorded at 0.0053 %. Inspired from the calculations, a multistage vacuum distillation method was successfully proposed and achieved to remove As from crude tin. Typically, the total material recovery rate is 99.96 %, and the direct rate of Sn is 93.02 %. The total power consumption and emission of this process are better than the corresponding parameters of other treatment processes, which is a promising solution for the cleaning and effective removal of As.

Self-Knowledge Distillation via Progressive Associative Learning

As a specific form of knowledge distillation (KD), self-knowledge distillation enables a student network to progressively distill its own knowledge without relying on a pretrained, complex teacher network; however, recent studies of self-KD have discovered that additional dark knowledge captured by auxiliary architecture or data augmentation could create better soft targets for enhancing the network but at the cost of significantly more computations and/or parameters. Moreover, most existing self-KD methods extract the soft label as a supervisory signal from individual input samples, which overlooks the knowledge of relationships among categories. Inspired by human associative learning, we propose a simple yet effective self-KD method named associative learning for self-distillation (ALSD), which progressively distills richer knowledge regarding the relationships between categories across independent samples. Specifically, in the process of distillation, the propagation of knowledge is weighted based on the intersample relationship between associated samples generated in different minibatches, which are progressively estimated with the current network. In this way, our ALSD framework achieves knowledge ensembling progressively across multiple samples using a single network, resulting in minimal computational and memory overhead compared to existing ensembling methods. Extensive experiments demonstrate that our ALSD method consistently boosts the classification performance of various architectures on multiple datasets. Notably, ALSD pushes forward the self-KD performance to 80.10% on CIFAR-100, which exceeds the standard backpropagation by 4.81%. Furthermore, we observe that the proposed method shows comparable performance with the state-of-the-art knowledge distillation methods without the pretrained teacher network.

Evaluation of the stability of hop pellets subjected to different storage conditions during three years

It is common knowledge that the composition of hops secondary metabolites depends on several factors, among which the storage conditions. In recent years, many studies have been conducted to evaluate changes in the chemical composition of hop products in response to different storage temperatures both in aerobic and anaerobic environments. However, little knowledge exists regarding the storage of hop pellets under modified atmosphere for a very long period of time. The aim of this study is to evaluate the effect on hop quality parameters of different storage conditions on U.S. hop cultivar pellets (Chinook and Nugget) in a three-year trial. Storage in modified atmosphere (atmosphere saturated with 70% nitrogen (N2) and 30% carbon dioxide (CO2) and vacuum storage were tested at two different temperatures, 4 °C and 10 °C. In order to establish the best storage condition, the bitter acids content (HPLC-UV method), the essential oil content (steam distillation method) and the essential oil profile (GC-MS method) were determined and characterised at 0, 6, 12 and 36 months. No statistical differences were observed between samples stored under different atmospheres, while significant differences were highlighted as a result of the storage temperature, e.g. the alpha acids content at 4 °C was significantly higher (P = 0.002). The two hop cultivars showed different behaviours in the storage trial, suggesting that high alpha hop varieties, like Nugget, could be stored even at higher temperatures.

Distillation of multi-class cervical lesion cell detection via synthesis-aided pre-training and patch-level feature alignment

Automated detection of cervical abnormal cells from Thin-prep cytologic test (TCT) images is crucial for efficient cervical abnormal screening using computer-aided diagnosis systems. However, the construction of the detection model is hindered by the preparation of the training images, which usually suffers from issues of class imbalance and incomplete annotations. Additionally, existing methods often overlook the visual feature correlations among cells, which are crucial in cervical lesion cell detection as pathologists commonly rely on surrounding cells for identification. In this paper, we propose a distillation framework that utilizes a patch-level pre-training network to guide the training of an image-level detection network, which can be applied to various detectors without changing their architectures during inference. The main contribution is three-fold: (1) We propose the Balanced Pre-training Model (BPM) as the patch-level cervical cell classification model, which employs an image synthesis model to construct a class-balanced patch dataset for pre-training. (2) We design the Score Correction Loss (SCL) to enable the detection network to distill knowledge from the BPM model, thereby mitigating the impact of incomplete annotations. (3) We design the Patch Correlation Consistency (PCC) strategy to exploit the correlation information of extracted cells, consistent with the behavior of cytopathologists. Experiments on public and private datasets demonstrate the superior performance of the proposed distillation method, as well as its adaptability to various detection architectures.

Unstructured Document Information Extraction Method with Multi-Faceted Domain Knowledge Graph Assistance for M2M Customs Risk Prevention and Screening Application

As a crucial national security defense line, the existing risk prevention and screening system of customs falls short in terms of intelligence and diversity for risk identification factors. Hence, the urgent issues to be addressed in the risk identification system include intelligent extraction technology for key information from Customs Unstructured Accompanying Documents (CUADs) and the reliability of the extraction results. In the customs scenario, OCR is employed for M2M interactions, but current models have difficulty adapting to diverse image qualities and complex customs document content. We propose a hybrid mutual learning knowledge distillation (HMLKD) method for optimizing a pre-trained OCR model’s performance against such challenges. Additionally, current models lack effective incorporation of domain-specific knowledge, resulting in insufficient text recognition accuracy for practical customs risk identification. We propose a customs domain knowledge graph (CDKG) developed using CUAD knowledge and propose an integrated CDKG post-OCR correction method (iCDKG-PostOCR) based on CDKG. The results on real data demonstrate that the accuracies improve for code text fields to 97.70%, for character type fields to 96.55%, and for numerical type fields to 96.00%, with a confidence rate exceeding 99% for each. Furthermore, the Customs Health Certificate Extraction System (CHCES) developed using the proposed method has been implemented and verified at Tianjin Customs in China, where it has showcased outstanding operational performance.

Feasibility of Different Methods for Separating n-Hexane and Ethanol

Conventional distillation methods cannot effectively separate the components of an azeotropic mixture since both phases have the same composition, thereby preventing further separation. Additional techniques such as pressure swing distillation or distillation with entrainers are often employed to overcome this limitation and achieve separation. The aim of this investigation was to select the most effective method for separating n-hexane and ethanol. The feasibility of three methods was analyzed: reduced pressure distillation, extractive distillation, and liquid–liquid extraction. The mutual solubility of n-hexane and prepared deep eutectic solvents (DESs) (nine hydrophilic: choline chloride with glycerol, ethylene glycol, or carboxylic acid (malic, citric, glycolic); tetramethylammonium chloride with glycolic acid; lactic acid with glycerol; K2CO3 with glycerol or ethylene glycol; two hydrophobic: menthol with decanoic or dodecanoic acid) was experimentally determined. Extraction experiments were conducted to test the solubility of DESs in the feed mixture. The effect of changing DES-to-feed mass ratio was further investigated with choline chloride–glycerol (1:2). The same DES and both hydrophobic DESs were able to increase the relative volatility and enhance the separation of ethanol and n-hexane. Based on the obtained results, extraction was selected as the most effective method for the separation of n-hexane and ethanol.

Preparation and modification of PVDF membrane via VIPS method for membrane distillation

Membrane distillation (MD) is a promising technology for treating high-salinity mine water. However, MD membranes are known to have low membrane flux and are prone to fouling. In this study, we used the cost-effective and controllable vapor-induced phase separation (VIPS) technology to prepare polyvinylidene difluoride (PVDF) membranes, replacing the traditional immersion precipitation method and optimizing the membrane structure by including LiCl and acetone as porogen in the casting solution. The results showed that the membrane prepared using the VIPS method exhibited a highly open interconnected porous surface. Unlike traditional MD membranes with a dense epidermal layer and large finger-like pores, these optimized membranes had a symmetrical and uniform internal structure, leading to a high flux of 8.62 kg·(m2·h)−1 during direct contact membrane distillation testing. Different porogens produced varied results on the VIPS process and varying effects on membrane structure. The use of LiCl promoted the formation of PVDF β-phase, resulting in a decrease in the number of spherical nodules on the membrane surface, as well as improved density and smoothness. Consequently, this reduced fouling risk during membrane distillation while slightly decreasing membrane flux. On the other hand, acetone rapidly evaporated during the VIPS process, facilitating pre-gelatinization and α-phase formation of PVDF. This concurrent effect effectively restricted excessive nodule growth on the membrane surface, endowing the membrane with antifouling capabilities while preserving high flux.