Conversion Model Research Articles

Exploring the role of process control and catalyst design in methane catalytic decomposition: A machine learning perspective

Methane catalytic pyrolysis presents a promising avenue for producing CO2-free hydrogen. However, optimizing catalyst composition and process parameters through experimental studies has been resource-intensive. Machine learning techniques have been increasingly utilized to deepen the understanding of process mechanisms. In this study, machine learning models were developed to deepen the understanding of the effects of process control and catalyst design on CH4 conversion performance. The optimal CH4 conversion model achieved a testing R2 of 0.9999. To further assess the model's real-world applicability, an initial verification experiment was conducted using a catalyst composition and process parameters distinct from those in the dataset. The results reaffirmed the model's good accuracy in real-world scenarios. Delving deeper into the intricacies of methane pyrolysis, the Shapley values and partial dependence plots generated by the model were scrutinized and discussed to provide a new perspective on the influence of process control and catalyst design parameters on methane catalytic decomposition.

Kinetic analysis and back propagation neural network model for shelf-life estimation of stabilized rice bran

To establish a general model for predicting shelf-life of stabilized rice bran, different samples with varying residual enzyme activity, water activity (aw), and packaging materials were used, and hot air assisted-radio frequency (HA-RF) heating was applied as stabilization method. Free fatty acids (FFA) were recorded during storage at different temperatures and kinetic analysis was conducted to compare the effects of different conditions on rice bran shelf-life. FFA content increased rapidly during storage, and higher aw and lipase (LA) activity corresponded to a higher FFA generation rate. Only the increase in FFA of samples with low LA activity (≤30 mg/g) and aw (≤0.3) conformed to Zero-order model, while that of other samples conformed to Fractional Conversion model. For the inhibitory effect of packaging materials on the generation of FFA, BOPP/Kraft-paper/PE > PA/PE > PET/PE > LDPE. A back propagation neural network (BPNN) model was also constructed to predict rice bran shelf-life, which employing LA activity, aw, packaging materials, and storage temperature as input parameters. The BPNN model trained by Levenberg–Marquardt achieved the best prediction performance, with coefficient of determination (R2) of 0.996 and root mean square error (RMSE) of 2.077. Rice bran packaged by BOPP/Kraft-paper/PE with LA activity of 20 mg/g and aw of 0.2 had the longest shelf-life of 147 d at 25 °C. This model is of great help in predicting the shelf-life of stabilized rice bran and guiding the development of rice bran stabilization process.

Active learning with biased non-response to label requests

AbstractActive learning can improve the efficiency of training prediction models by identifying the most informative new labels to acquire. However, non-response to label requests can impact active learning’s effectiveness in real-world contexts. We conceptualise this degradation by considering the type of non-response present in the data, demonstrating that biased non-response is particularly detrimental to model performance. We argue that biased non-response is likely in contexts where the labelling process, by nature, relies on user interactions. To mitigate the impact of biased non-response, we propose a cost-based correction to the sampling strategy–the Upper Confidence Bound of the Expected Utility (UCB-EU)–that can, plausibly, be applied to any active learning algorithm. Through experiments, we demonstrate that our method successfully reduces the harm from labelling non-response in many settings. However, we also characterise settings where the non-response bias in the annotations remains detrimental under UCB-EU for specific sampling methods and data generating processes. Finally, we evaluate our method on a real-world dataset from an e-commerce platform. We show that UCB-EU yields substantial performance improvements to conversion models that are trained on clicked impressions. Most generally, this research serves to both better conceptualise the interplay between types of non-response and model improvements via active learning, and to provide a practical, easy-to-implement correction that mitigates model degradation.

Conversion Design of Arduino-Based Semi-Automatic Clothes Washer

This research aims to develop and evaluate a system that converts a manual clothes washer into an Arduino-based semi-automatic one. The research method used is a method of designing and creating a system for the object of research or called the design method. The system design is used as an object where the system is discussed and tested to find out how the system works and performs itself. The design object is an efficient and practical clothes washing tool.Converting a manual clothes washer into an Arduino-based semi-automatic one is an innovative and efficient solution to increase convenience and effectiveness in the clothes washing process. The design of this conversion model provides a complete guide to convert a manual washing machine into a semi-automatic one, improving the efficiency and convenience of daily use. With proper implementation, this project can provide an effective and cost-effective solution for household users.

Incorporating low haemoglobin into a risk prediction model for conversion in minimally invasive gynaecologic oncology surgeries

Background A well-known complication of laparoscopic management of gynaecologic masses and cancers is the need to perform an intraoperative conversion to laparotomy. The purpose of this study was to identify novel patient risk factors for conversion from minimally invasive to open surgeries for gynaecologic oncology operations. Methods This was a retrospective cohort study of 1356 patients ≥18 years of age who underwent surgeries for gynaecologic masses or malignancies between February 2015 and May 2020 at a single academic medical centre. Multivariable logistic regression was used to study the effects of older age, higher body mass index (BMI), higher American Society of Anaesthesiologist (ASA) physical status, and lower preoperative haemoglobin (Hb) on odds of converting from minimally invasive to open surgery. Receiver operating characteristic (ROC) curve analysis assessed the discriminatory ability of a risk prediction model for conversion. Results A total of 704 planned minimally invasive surgeries were included with an overall conversion rate of 6.1% (43/704). Preoperative Hb was lowest for conversion cases, compared to minimally invasive and open cases (11.6 ± 1.9 vs 12.8 ± 1.5 vs 11.8 ± 1.9 g/dL, p<.001). Patients with preoperative Hb <10 g/dL had an adjusted odds ratio (OR) of 3.94 (CI: 1.65–9.41, p=.002) for conversion while patients with BMI ≥30 kg/m2 had an adjusted OR of 2.86 (CI: 1.50–5.46, p=.001) for conversion. ROC curve analysis using predictive variables of age >50 years, BMI ≥30 kg/m2, ASA physical status >2, and preoperative haemoglobin <10 g/dL resulted in an area under the ROC curve of 0.71. Patients with 2 or more risk factors were at highest risk of requiring an intraoperative conversion (12.0%). Conclusions Lower preoperative haemoglobin is a novel risk factor for conversion from minimally invasive to open gynaecologic oncology surgeries and stratifying patients based on conversion risk may be helpful for preoperative planning.

Spatial utilization of historical topographic map and its application in land reconstruction of ancient Chinese urban land use

The historical topographic map preserves rich geographic information and can provide direct assistance for the reconstruction of various geographic elements. Based on the historical data of cities throughout the Qing Dynasty, the land use scale data of cities across the country was obtained using GIS and urban perimeter conversion models. This study combines city information and city circumference records from the historical maps and archives of the late Qing Dynasty to quantitatively reconstruct the use patterns of ancient China’s urban land at a spatial resolution of 1° × 1°. Uncertainty analysis of the reconstruction results was conducted using modern remote sensing image data as the validation data set. The results showed the following. (1) During the late Qing Dynasty, the total area of urban land in the various provinces and regions was 1456.015 km2. The maximum value was 208.691 km2 in Beijing-Tianjin-Hebei, the minimum value was 1.713 km2 in Qinghai, and the average value was 56.001 km2. (2) The results of grid reconstruction show that among the 398 grids with urban land distribution, the maximum value is 64.099 km2/grid, the minimum value is 0.013 km2/grid, and the average value is 3.658 km2/grid. (3) Of all the grids with urban land, the urban land grid to the west of the Hu Line accounts for 12.5% and the east to 87.5%. (4) During the late Qing Dynasty, urban land use in China was primarily concentrated in agriculturally developed areas such as the North China Plain, the Central Plains, Jiangnan, and the Sichuan-Chongqing region. (6) The results of a kernel density estimation showed that there were obviously three core areas of urban land agglomeration in China during the late Qing Dynasty: the North China Plain-Central Plains, the Jiangsu-Shanghai-Zhejiang-Anhui area, and the Sichuan-Chongqing urban core area. This study provides basic data for urban land use during historical periods and provides a basis for the quantitative reconstruction of relevant urban land data for historical archives.

A grid model of direct conversion between zenith tropospheric delay and precipitable water vapor in tropical regions

A grid model of direct conversion between zenith tropospheric delay and precipitable water vapor in tropical regions

MOSS: An Open Conversational Large Language Model

Conversational large language models (LLMs) such as ChatGPT and GPT-4 have recently exhibited remarkable capabilities across various domains, capturing widespread attention from the public. To facilitate this line of research, in this paper, we report the development of MOSS, an open-sourced conversational LLM that contains 16 B parameters and can perform a variety of instructions in multi-turn interactions with humans. The base model of MOSS is pre-trained on large-scale unlabeled English, Chinese, and code data. To optimize the model for dialogue, we generate 1.1 M synthetic conversations based on user prompts collected through our earlier versions of the model API. We then perform preference-aware training on preference data annotated from AI feedback. Evaluation results on real-world use cases and academic benchmarks demonstrate the effectiveness of the proposed approaches. In addition, we present an effective practice to augment MOSS with several external tools. Through the development of MOSS, we have established a complete technical roadmap for large language models from pre-training, supervised fine-tuning to alignment, verifying the feasibility of chatGPT under resource-limited conditions and providing a reference for both the academic and industrial communities. Model weights and code are publicly available at https://github.com/OpenMOSS/MOSS.

Opinion Identification using a Conversational Large Language Mode

The paper focuses on testing the use of conversational Large Language Model (LLM), in particularchatGPT and Google models, instructed to assume the role of linguistics experts to produce opinions. Incontrast to knowledge/evidence-based objective factual statements, opinions are defined as subjective statements about animates, things, events or properties in the context of an Opinion (Speech) Event in a social cultural context. Taxonomy distinguishes explicit (direct/indirect) and implicit opinions (positive, negative, ambiguous, or balanced). Contextually richer prompts at the LLMs training phase are shown to be needed to deal with variants of implicit opinion scenario types.

Paired plasma lipidomics and proteomics analysis in the conversion from mild cognitive impairment to Alzheimer's disease

BackgroundAlzheimer's disease (AD) is a neurodegenerative condition for which there is currently no available medication that can stop its progression. Previous studies suggest that mild cognitive impairment (MCI) is a phase that precedes the disease. Therefore, a better understanding of the molecular mechanisms behind MCI conversion to AD is needed. MethodHere, we propose a machine learning-based approach to detect the key metabolites and proteins involved in MCI progression to AD using data from the European Medical Information Framework for Alzheimer's Disease Multimodal Biomarker Discovery Study. Proteins and metabolites were evaluated separately in multiclass models (controls, MCI and AD) and together in MCI conversion models (MCI stable vs converter). Only features selected as relevant by 3/4 algorithms proposed were kept for downstream analysis. ResultsMulticlass models of metabolites highlighted nine features further validated in an independent cohort (0.726 mean balanced accuracy). Among these features, one metabolite, oleamide, was selected by all the algorithms. Further in-vitro experiments in rodents showed that disease-associated microglia excreted oleamide in vesicles. Multiclass models of proteins stood out with nine features, validated in an independent cohort (0.720 mean balanced accuracy). However, none of the proteins was selected by all the algorithms. Besides, to distinguish between MCI stable and converters, 14 key features were selected (0.872 AUC), including tTau, alpha-synuclein (SNCA), junctophilin-3 (JPH3), properdin (CFP) and peptidase inhibitor 15 (PI15) among others. ConclusionsThis omics integration approach highlighted a set of molecules associated with MCI conversion important in neuronal and glia inflammation pathways.

Application of MR images in radiotherapy planning for brain tumor based on deep learning

Purpose Explore the function and dose calculation accuracy of MRI images in radiotherapy planning through deep learning methods. Methods 131 brain tumor patients undergoing radiotherapy with previous MR and CT images were recruited for this study. A new series of MRI from the aligned MR was firstly registered to CT images strictly using MIM software and then resampled. A deep learning method (U-NET) was used to establish a MRI-to-CT conversion model, for which 105 patient images were used as the training set and 26 patient images were used as the tuning set. Data from additional 8 patients were collected as the test set, and the accuracy of the model was evaluated from a dosimetric standpoint. Results Comparing the synthetic CT images with the original CT images, the difference in dosimetric parameters D98, D95, D2 and Dmean of PTV in 8 patients was less than 0.5%. The gamma passed rates of PTV and whole body volume were: 1%/1 mm: 93.96%±6.75%, 2%/2 mm: 99.87%±0.30%, 3%/3 mm: 100.00%±0.00%; and 1%/1 mm: 99.14%±0.80%, 2%/2 mm: 99.92%±0.08%, 3%/3 mm: 99.99%±0.01%. Conclusion MR images can be used both in delineation and treatment efficacy evaluation and in dose calculation. Using the deep learning way to convert MR image to CT image is a viable method and can be further used in dose calculation.

On the inaccuracies of point-particle approach for char conversion modeling

Char conversion is a complex phenomenon that involves not only heterogeneous reactions but also external and internal heat and mass transfer. Reactor-scale simulations often use a point-particle approach (PP approach) as sub-models for char conversion because of its low computational cost. Despite a number of simplifications involved in the PP approach, there are very few studies that systematically investigate the inaccuracies of the PP approach. This study aims to compare and identify when and why the PP approach deviates from resolved-particle simulations (RP approach). Simulations have been carried out for CO2 gasification of a char particle under zone II conditions (i.e., pore diffusion control) using both PP and RP approaches. Results showed significant deviations between the two approaches for the effectiveness factor, gas compositions, particle temperature, and particle diameter. The most significant sources of inaccuracies in the PP approach are negligence of the non-uniform temperature inside the particle and the inability to accurately model external heat transfer. Under the conditions with low effectiveness factors, the errors of intra-particle processes were dominant while the errors of external processes became dominant when effectiveness factors were close to unity. Because it assumes uniform internal temperature, the models applying the PP approach always predict higher effectiveness factors than the RP approach, despite its accurate estimation of intra-particle mass diffusion effects. As a consequence, the PP approach failed to predict the particle size changes accurately. Meanwhile, no conventional term for external heat transfer could explain the inaccuracy, indicating the importance of other sources of errors such as 2D/3D asymmetry or penetration of external flows inside the particles.

ANN based LIBS models for quasi-experimental spectra relevant for materials for next-step fusion reactors

Following the successful demonstration of machine learning (ML) models for laser induced breakdown spectroscopy (LIBS) adaptation in fusion reactor fuel retention monitoring using synthetic data [Gąsior et al., Spectrochim. Acta, Part B 199, 106576 (2023)], this study focuses on implementing operability on experimental data. To achieve this, Simulated Eperimental Spectra (SES) data are generated and used for validation of a chemical composition estimation model trained on dimensionally reduced synthetic spectral data (DRSSD). Principal component analysis is employed for dimensionality reduction of both SES and DRSSD. To simulate real experimental conditions, the synthetic data, generated by a dedicated tool [M. Kastek (2022), “SimulatedLIBS,” Zenodo. http://dx.doi.org/10.5281/zenodo.7369805] is processed through the transmission function of a real spectroscopy setup at IPPLM. Separate and optimized artificial neural network models are implemented for conversion and chemical composition estimation. The conversion model takes DR-SES as features and DR-SSD as targets. Validation using converted SES data demonstrates chemical composition predictions comparable to those from synthetic data, with the highest relative uncertainty increase below 40% and a normalized root-mean-square error of prediction below 7%. This work represents a significant step toward adapting ML-based LIBS for fuel and impurity retention monitoring in the walls of next-generation fusion devices.

High-performance deep spiking neural networks via at-most-two-spike exponential coding

Spiking neural networks (SNNs) provide necessary models and algorithms for neuromorphic computing. A popular way of building high-performance deep SNNs is to convert ANNs to SNNs, taking advantage of advanced and well-trained ANNs. Here we propose an ANN to SNN conversion methodology that uses a time-based coding scheme, named At-most-two-spike Exponential Coding (AEC), and a corresponding AEC spiking neuron model for ANN-SNN conversion. AEC neurons employ quantization-compensating spikes to improve coding accuracy and capacity, with each neuron generating up to two spikes within the time window. Two exponential decay functions with tunable parameters are proposed to represent the dynamic encoding thresholds, based on which pixel intensities are encoded into spike times and spike times are decoded into pixel intensities. The hyper-parameters of AEC neurons are fine-tuned based on the loss function of SNN-decoded values and ANN-activation values. In addition, we design two regularization terms for the number of spikes, providing the possibility to achieve the best trade-off between accuracy, latency and power consumption. The experimental results show that, compared to other similar methods, the proposed scheme not only obtains deep SNNs with higher accuracy, but also has more significant advantages in terms of energy efficiency and inference latency. More details can be found at https://github.com/RPDS2020/AEC.git.

Comprehensive investigation of almond shells pyrolysis using advance predictive models

This research focused on comprehensive characterization and assessment of almond shells pyrolysis for bioenergy potential through thermogravimetric analysis from ambient temperature to 900 °C at different heating rates of 10, 15, and 20 °C/min in inert environment. Iso-conversional model-free methods like Friedman, Ozawa-Flynn-Wall (OFW), and Kissinger-Akahira-Sunose (KAS) were used for kinetic analysis. Average activation energies (Ea) evaluated using Friedman, OFW, and KAS methods were 198.45 kJ mol−1, 204.43 kJ mol−1, and 204.97 kJ mol−1, respectively. The evaluation of thermodynamic parameters, including ΔH‡, ΔG‡, and ΔS‡, was also assessed. The average values of ΔH‡, ΔG‡, and ΔS‡, were found to be 199.4 kJ mol−1, 172.17 kJ mol−1 and 42.60 kJ mol−1 respectively. The reaction mechanism was obtained from combined kinetics. A high R2 value of 0.9933 demonstrates strong agreement between the combined kinetic analysis results and the experimental data. The distribution activation energy model was assessed employing four pseudo elements identified as PC1, PC2, PC3, and PC4. Artificial Neural Network (ANN) and Boosting regression trees (BRT) were used for the prediction of Ea of almond shells pyrolysis. The detailed understanding of thermokinetics and creating customized predictive and innovative modelling techniques like ANN and BRT sets a new benchmark for developing customized models for thermochemical conversion of varieties of almond shells.

Robust 3D coordinate transformation based on genetic algorithm intelligent weighting

In order to improve the solution quality of 3D coordinate transformation parameters, a robust optimization model is obtained by integrating the 3D coordinate transformation model within the overall framework of the genetic algorithm in this paper. The minimum of the weighted sum of squares of the point residuals of the common points is taken as the objective function for optimization, and the parallel search is started from the initial population, and the updating process of the solution set is used to generate new and better individuals, and the adaptive weight combination corresponding to the optimum of the fitness function is obtained at the end of the generation iteration. The generated simulated data and measured data were analyzed using the Hefei Light Source tunnel control network as an application scenario. The results show that the method in this paper can automatically reduce the influence of low-quality data on the model, and compared with the two robust conversion models of IGG III iterative weighting, the method in this paper is not affected by the empirical parameters in determining the weight combinations and the weighted sum of squares of the residuals of the solved points are smaller, so it can effectively improve the quality of the solved conversion parameters.

Grammar, cohesion and the co-ordination of the “self” in a current psychotherapeutic technique

Abstract This paper aims to show how concepts and analytical methods of systemic functional linguistics can work congruently with other human practices to improve outcomes for those undergoing the suffering around loss of meaning and the absence of purposeful, self-directed experience. Based on a two-decade collaboration between linguists and psychotherapists in Sydney, Australia, and using the tools of text linguistics as developed by Michael A. K. Halliday and Ruqaiya Hasan in systemic functional theory, the paper presents an indicative selection of intense exchanges between traumatized persons and therapists (centrally the experience of ‘Ruth’). The level by level linguistic descriptions of these exchanges offer opportunities for understanding how progress in the clinical interaction might be achieved. The descriptions can also be evaluated against the theoretical claims of psychotherapy in psychiatry – in particular, the emphasis of the Conversational Model of Psychotherapy developed in England and Australia by Robert Hobson and Russell Meares, whose characterization of disorders involves an emphasis on ‘co-ordination’ and ‘cohesion’ within frontal lobe activity of traumatized patients. In this way the paper also explores conceptual parallels and intellectual antecedents shared between the Conversational Model and Systemic Functional Linguistics, contributing to the broader intellectual history of the human sciences.

Variant.ai

This research paper presents a novel approach to developing a Software as a Service (SaaS) AI platform aimed at revolutionizing user interactions through dynamic conversations facilitated by an advanced AI conversation model. The platform prioritizes user-centric design, leveraging cutting-edge technologies to ensure a seamless and visually engaging experience. Key features include an emphasis on visual excellence through meticulous UI design and animations, as well as robust data integrity measures implemented for secure user data management. The platform also emphasizes operational resilience, with effective server error handling mechanisms in place to enhance user experience. Additionally, the integration of a Conversation Generation Tool, powered by advanced AI models, further enhances user engagement by enabling interactive and dynamic conversations. This research contributes to the advancement of SaaS platforms by demonstrating innovative approaches to user-centric design and AI integration. Key Words: Software as a Service (SaaS), Dynamic conversations, Advanced AI models, Conversation Generation Tool

Enhancing Voice Assistants with Generative AI Language Models for Interactive Conversations

This paper presents comprehensive research for the development of a voice assistant application using HTML/CSS, JavaScript, Python, SQLite, and the Hugging Face API. The proposed methodology encompasses key steps in data collection, model training, frontend and backend development, integration, testing, and deployment, ensuring the creation of a robust and user-friendly voice assistant system. By leveraging a combination of web technologies and Al tools, the developed voice assistant application offers an interactive and intelligent user experience across its users. Key Words: Voice assistant, HTML/CSS, JavaScript, Python, SQLite, Hugging Face API, LLMs, AI.

Understanding the Coupling Mechanism of Intercalation and Conversion Hybrid Storage in Lithium-Graphite Anode.

Anodes with high capacity and long lifespan play an important role in the advanced batteries. However, none of the existing anodes can meet these two requirements simultaneously. Lithium (Li)-graphite composite anode presents great potential in balancing these two requirements. Herein, the working mechanism of Li-graphite composite anode is comprehensively investigated. The capacity decay features of the composite anode are different from those of Li ion intercalation in Li ion batteries and Li metal deposition in Li metal batteries. An intercalation and conversion hybrid storage mechanism are proposed by analyzing the capacity decay ratios in the composite anode with different initial specific capacities. The capacity decay models can be divided into four stages including Capacity Retention Stage, Relatively Independent Operation Stage, Intercalation & Conversion Coupling Stage, Pure Li Intercalation Stage. When the specific capacity is between 340 and 450mAh g-1, its capacity decay ratio is between that of pure intercalation and conversion model. These results intensify the comprehensive understandings on the working principles in Li-graphite composite anode and present novel insights in the design of high-capacity and long-lifespan anode materials for the next-generation batteries.