Fine-tuning Of Parameters Research Articles

Process Development for Newcastle Disease Virus-Vectored Vaccines in Serum-Free Vero Cell Suspension Cultures.

The ongoing COVID-19 pandemic drew global attention to infectious diseases, attracting numerous resources for development of pandemic preparedness plans and vaccine platforms—technologies with robust manufacturing processes that can quickly be pivoted to target emerging diseases. Newcastle Disease Virus (NDV) has been studied as a viral vector for human and veterinary vaccines, but its production relies heavily on embryonated chicken eggs, with very few studies producing NDV in cell culture. Here, NDV is produced in suspension Vero cells, and analytical assays (TCID50 and ddPCR) are developed to quantify infectious and total viral titer. NDV-GFP and NDV-FLS (SARS-CoV-2 full-length spike protein) constructs were adapted to replicate in Vero and HEK293 suspension cultures using serum-free media, while fine-tuning parameters such as MOI, temperature, and trypsin concentration. Shake flask productions with Vero cells resulted in infectious titers of 1.07 × 108 TCID50/mL for NDV-GFP and 1.33 × 108 TCID50/mL for NDV-FLS. Production in 1 L batch bioreactors also resulted in high titers in culture supernatants, reaching 2.37 × 108 TCID50/mL for NDV-GFP and 3.16 × 107 TCID50/mL for NDV-FLS. This shows effective NDV production in cell culture, building the basis for a scalable vectored-vaccine manufacturing process that can be applied to different targets.

Realistic simulation and real-time reconstruction of digital holographic microscopy experiments in ImageJ.

The description, implementation, and validation of an ImageJ plugin that allows the realistic simulation and real-time reconstruction of digital holographic microscopy (DHM) experiments are presented. The simulation module implements a telecentric image-plane DHM recording scheme with fully configurable imaging system, interference, and scaling parameters, including the possibility of defining an estimate of the roughness distribution of the sample to produce realistic coherent-noise affectations. The reconstruction module allows the computation of amplitude, intensity, or phase, from digital holograms' input as either single images or video streams for real-time processing; this module also implements user-defined fine-tuning parameters, allowing subpixel linear phase compensations and digital refocusing of the complex-valued reconstructed fields. In this note, the functionality of the plugin is illustrated by simulating the noisy DHM recording of a phase-only resolution test target and the reconstruction of both the resulting synthetic hologram and an equivalent experimental recording; the results show good agreement between the simulation and the experimental recording, and accurate measurements on the reconstructed information, thus granting the use of either module with full confidence according to needs and possibilities.

Research on fine tuning parameter insulator identification based on YOLOV4 algorithm

Aiming at the problem that the training network time of YOLOV4 algorithm is too long due to the large data set of aerial insulator images, a method based on YOLOV4 algorithm is proposed to shorten the training time by fine-tuning parameters without affecting the positioning detection accuracy. Based on the development of UBANTU virtual machine, through CUDA and CUDNN environment configuration, and through the detection and verification of insulator aerial photo data set, the feasibility of accurate positioning of insulators under the condition of fine tuning parameters of YOLOV4 algorithm is successfully proved.

Unravelling the anaerobic digestion ‘black box’: Biotechnological approaches for process optimization

Increasing global demand for green energy has necessitated more research interest in anaerobic digestion. Due to the dual benefits of anaerobic digestion, in organic waste management and energy generation, it is rated among the best renewable energy generation options. An important challenge that limits acceptability and adoption of this technology is the quality and quantity of anaerobic digestion products, especially biogas and digestate. A practical solution to this challenge is process optimization. However, process optimization is not easily achieved due to the complexity of the microbiome driving the anaerobic digestion process, which has contributed to anaerobic digesters being considered ‘black boxes’. Unravelling the ‘black box’ would aid in fine-tuning operational conditions, process parameters, intra- and inter-phase interactions as well as rate limiting factors during each phase of the production cycle, thereby optimizing the anaerobic digestion process. To achieve this level of technological advancement, cooperation between experts that understand both the microbial processes and technological design of digesters is important. When factors are optimized, implementation of the optimized conditions through automation may prevent process errors and failures, thereby enhancing quality and quantity of anaerobic digestion products. In addition, process optimization should also reflect economic viability vis-à-vis small-scale and large-scale biogas production systems. This review provides a detailed overview of individual factors that could affect the performance of the anaerobic digestion process with critical interpretations and suggestions. This work could be employed as a good base to build integrative models or ecological indices that interrelate these factors.

Runtime Analyses of the Population-Based Univariate Estimation of Distribution Algorithms on LeadingOnes

We perform rigorous runtime analyses for the univariate marginal distribution algorithm (UMDA) and the population-based incremental learning (PBIL) Algorithm on LeadingOnes. For the UMDA, the currently known expected runtime on the function is {mathcal {O}}left( nlambda log lambda +n^2right) under an offspring population size lambda =Omega (log n) and a parent population size mu le lambda /(e(1+delta )) for any constant delta >0 (Dang and Lehre, GECCO 2015). There is no lower bound on the expected runtime under the same parameter settings. It also remains unknown whether the algorithm can still optimise the LeadingOnes function within a polynomial runtime when mu ge lambda /(e(1+delta )). In case of the PBIL, an expected runtime of {mathcal {O}}(n^{2+c}) holds for some constant c in (0,1) (Wu, Kolonko and Möhring, IEEE TEVC 2017). Despite being a generalisation of the UMDA, this upper bound is significantly asymptotically looser than the upper bound of {mathcal {O}}left( n^2right) of the UMDA for lambda =Omega (log n)cap {mathcal {O}}left( n/log nright). Furthermore, the required population size is very large, i.e., lambda =Omega (n^{1+c}). Our contributions are then threefold: (1) we show that the UMDA with mu =Omega (log n) and lambda le mu e^{1-varepsilon }/(1+delta ) for any constants varepsilon in (0,1) and 0<delta le e^{1-varepsilon }-1 requires an expected runtime of e^{Omega (mu )} on LeadingOnes, (2) an upper bound of {mathcal {O}}left( nlambda log lambda +n^2right) is shown for the PBIL, which improves the current bound {mathcal {O}}left( n^{2+c}right) by a significant factor of Theta (n^{c}), and (3) we for the first time consider the two algorithms on the LeadingOnes function in a noisy environment and obtain an expected runtime of {mathcal {O}}left( n^2right) for appropriate parameter settings. Our results emphasise that despite the independence assumption in the probabilistic models, the UMDA and the PBIL with fine-tuned parameter choices can still cope very well with variable interactions.

Linking event archives to news: a computational method for analyzing the gatekeeping process

ABSTRACT Digital archives that document real-world events provide new opportunities for large-scale analyses of how news coverage represents reality. We present a method and open-source tool for linking event data to news articles, and demonstrate its application with an analysis of event and country level predictors of terrorism coverage in The Guardian from 2006 to 2018, using event data from the Global Terrorism Database (GTD). Our method builds on established techniques for calculating document similarity, and we propose a novel strategy for fine-tuning parameters of the event matching algorithm that requires no manual coding. An online appendix is provided that documents all code to replicate our analysis and reuse our tools.

Charged black hole in 4D Einstein-Gauss-Bonnet gravity: particle motion, plasma effect on weak gravitational lensing and centre-of-mass energy

We study the motion of charged and spinning particles and photons in the 4D charged Einstein-Gauss-Bonnet (EGB) black hole vicinity. We determine the radius of the innermost stable circular orbit (ISCO) for test particles. We show that the combined effect of the Gauss-Bonnet (GB) coupling parameter and black hole charge decreases the ISCO and the radius of the photon sphere. Further, we study the gravitational deflection angle and show that the impact of GB term and black hole charge on it is quite noticeable. We also consider the effect of plasma and find the analytical form of the deflection angle in the case of a uniform and non-uniform plasma. Interestingly we find that the deflection angle becomes larger when uniform plasma is considered in comparison to the case of non-uniform plasma. We also study the center of mass energy (E C.M.) obtained by collision process for non-spinning particles and show that the impact of GB parameter and black hole charge leads to high energy collision. In addition, we also study the E C.M. for the case of spinning particles and show that if the two spinning particles collide near the horizon of 4D charged EGB BH, the E C.M. becomes infinitely high which is in disparity with the non-spinning particles counterpart where E C.M. never grows infinitely. To achieve this, an important role is played by the spinning particle known as the near-critical particle (i.e. a particle with fine-tuned parameters). In order to achieve the unbounded E C.M. from the collision of two spinning particles, the energy per unit mass must be less than unity for a near-critical particle, which means such a particle starts from some intermediate position r > rh and not from infinity.

Characters, Spirals and Unit Dynamics by training neural network model

Deep learning is an integral part of machine learning by building a structure like the neural network in the human brain. Amounts of different neural network models have been emerged, improved, and transformed. This article will implement various neural network models for three different specific tasks (Japanese character recognition/double helix task/remote unit dynamics) and analyse the fine-tuning parameters process (different network layers, different types of network layers). The result is that generally, models with more layers would perform better. However, other parameter could have significant effects on the performance. For example, activation function, different numbers of hidden nodes, different initial weights, and different periods, etc.

Enhanced optimization assisted interference mitigation with vertical beamforming in 3D MIMO-OFDMA for 5G wireless communication network

PurposeMultiple input multiple-output (MIMO) has emerged as one among the many noteworthy technologies in recent wireless applications because of its powerful ability to improve bandwidth efficiency and performance, i.e. through developing its unique spatial multiplexing capability and spatial diversity gain. For carrying out an enhanced communication in next-generation networks, the MIMO and orthogonal frequency division multiple systems were combined that facilitate the spatial multiplexing on resource blocks (RBs) based on time-frequency. This paper aims to propose a novel approach for maximizing the throughput of cell-edge users and cell-center users.Design/methodology/approachIn this work, the specified multi-objective function is defined as the single objective function, which is solved by the introduction of a new improved algorithm as well. This optimization problem can be resolved by the fine-tuning of certain parameters such as assigned power for RB, cell-center user, cell-edge user and RB allocation. The fine-tuning of parameters is attained by a new improved Lion algorithm (LA), termed as Lion with new cub generation (LA-NCG) model. Finally, the betterment of the presented approach is validated over the existing models in terms of signal to interference plus noise ratio, throughput and so on.FindingsOn examining the outputs, the adopted LA-NCG model for 4BS was 66.67%, 66.67% and 20% superior to existing joint processing coordinated multiple point-based dual decomposition method (JC-DDM), fractional programming (FP) and LA models. In addition, the throughput of conventional JC-DDM, FP and LA models lie at a range of 10, 45 and 35, respectively, at the 100th iteration. However, the presented LA-NCG scheme accomplishes a higher throughput of 58. Similarly, the throughput of the adopted scheme observed for 8BS was 59.68%, 44.19% and 9.68% superior to existing JC-DDM, FP and LA models. Thus, the enhancement of the adopted LA-NCG model has been validated effectively from the attained outcomes.Originality/valueThis paper adopts the latest optimization algorithm called LA-NCG to establish a novel approach for maximizing the throughput of cell-edge users and cell-center users. This is the first that work uses LA-NCG-based optimization that assists in fine-tuning certain parameters such as assigned power for RB, cell-center user, cell-edge user and RB allocation.

Fine-Tuning of Process Parameters Modulates Specific Metabolic Bacterial Activities and Aroma Compound Production in Semi-Hard Cheese.

The formation of cheese flavor mainly results from the production of volatile compounds by microorganisms. We investigated how fine-tuning cheese-making process parameters changed the cheese volatilome in a semi-hard cheese inoculated with Lactococcus (L.) lactis, Lactiplantibacillus (L.) plantarum, and Propionibacterium (P.) freudenreichii. A standard (Std) cheese was compared with three variants of technological itineraries: a shorter salting time (7 h vs 10 h, Salt7h), a shorter stirring time (15 min vs 30 min, Stir15min), or a higher ripening temperature (16 °C vs 13 °C, Rip16°C). Bacterial counts were similar in the four cheese types, except for a 1.4 log10 reduction of L. lactis counts in Rip16°C cheeses after 7 weeks of ripening. Compared to Std, Stir15min and Rip16°C increased propionibacterial activity, causing higher concentrations of acetic, succinic, and propanoic acids and lower levels of lactic acid. Rip16°C accelerated secondary proteolysis and volatile production. We thus demonstrated that fine-tuning process parameters could modulate the cheese volatilome by influencing specific bacterial metabolisms.

Raman micro-spectroscopic map estimating in vivo precision of tumor ablative effect achieved by photothermal therapy procedure

Photothermal-therapy (PTT) inculcates near-infrared laser guided local heating effect, where high degree of precision is expected, but not well proven to-date. An ex vivo tissue biochemical map with molecular/biochemical response showing the coverage area out of an optimized PTT procedure can reveal precision information. In this work, Raman-microscopic mapping and linear discriminant analysis of spectra of PTT treated and surrounding tissue areas ex vivo are done, revealing three distinct spectral clusters/zones, with minimal overlap between the core treated and adjacent untreated zone. The core treated zone showed intense nucleic-acid, cytochrome/mitochondria and protein damage, an adjacent zone showed lesser degree of damages and far zone showed minimal/no damage. Immunohistochemistry for γH2AX (DNA damage marker protein) in PTT exposed tissue also revealed similar results. Altogether, this study reveals the utility of Raman-microspectroscopy for fine-tuning safety parameters and precision that can be achieved from PTT mediated tumor ablation in preclinical/clinical application.

Anomaly Detection Using a Sliding Window Technique and Data Imputation with Machine Learning for Hydrological Time Series

Water level data obtained from telemetry stations typically contains large number of outliers. Anomaly detection and a data imputation are necessary steps in a data monitoring system. Anomaly data can be detected if its values lie outside of a normal pattern distribution. We developed a median-based statistical outlier detection approach using a sliding window technique. In order to fill anomalies, various interpolation techniques were considered. Our proposed framework exhibited promising results after evaluating with F1-score and root mean square error (RMSE) based on our artificially induced data points. The present system can also be easily applied to various patterns of hydrological time series with diverse choices of internal methods and fine-tuned parameters. Specifically, the Spline interpolation method yielded a superior performance on non-cyclical data while the long short-term memory (LSTM) outperformed other interpolation methods on a distinct tidal data pattern.

Electrochemical additive manufacturing of graded NiCoFeCu structures for electromagnetic applications

This study introduces the concept of compositionally-graded electrochemical additive manufacturing of advanced metal alloys. The feasibility was demonstrated using a NiCoFeCu system for electromagnetic applications. The effects of grading frequency and voltage on output geometry and elemental composition were studied. Application of constant signals throughout pillar build yielded varying width and elemental composition along the pillar length. Additional grading of voltage from low to high favoured wider pillars and enhanced copper content. Auxiliary plating effects were used to explain compositional deviations from conventional electroplating. Frequency response varied with voltage, suggesting its use as a secondary, fine-tuned parameter.

Cybersecurity Standards in the Context of Operating System

Cyber threats have been growing tremendously in recent years. There are significant advancements in the threat space that have led towards an essential need for the strengthening of digital infrastructure security. Better security can be achieved by fine-tuning system parameters to the best and optimized security levels. For the protection of infrastructure and information systems, several guidelines have been provided by well-known organizations in the form of cybersecurity standards. Since security vulnerabilities incur a very high degree of financial, reputational, informational, and organizational security compromise, it is imperative that a baseline for standard compliance be established. The selection of security standards and extracting requirements from those standards in an organizational context is a tedious task. This article presents a detailed literature review, a comprehensive analysis of various cybersecurity standards, and statistics of cyber-attacks related to operating systems (OS). In addition to that, an explicit comparison between the frameworks, tools, and software available for OS compliance testing is provided. An in-depth analysis of the most common software solutions ensuring compliance with certain cybersecurity standards is also presented. Finally, based on the cybersecurity standards under consideration, a comprehensive set of minimum requirements is proposed for OS hardening and a few open research challenges are discussed.

Determination of Elemental Composition in Soft Biological Tissue Using Laser Ablation Inductively Coupled Plasma Mass Spectrometry: Method Validation.

Determination of elemental concentrations in biological tissue is fundamental to many environmental studies. Analytical methods typically used to quantify concentrations in such studies have minimum sample volumes that necessitate lethal or impactful collection of tissues. Laser ablation inductively coupled mass spectrometry (LA-ICP-MS) has small sample volume requirements and offers environmental practitioners an opportunity to employ low-impact sample collection methods. Environmental applications of LA-ICP-MS are limited by the lack of validated methods, partly due to the need for dry samples and scarcity of matrix-matched certified reference materials (CRMs). This study validates an LA-ICP-MS method to determine concentrations of 30 elements in soft biological tissue (fish ovary and muscle). Tissue samples (median: 0.48 grams (g); inter-quartile range: 0.30 g to 0.56 g wet weight) were dehydrated, powdered, compressed into pellets (weighing approximately 0.03 g) and analyzed using LA-ICP-MS alongside three matrix-matched CRMs. The method yielded concentration determinations for CRM elements that were typically accurate to within 30% of theoretical concentrations, and precise (relative standard deviation <20%). These results were repeatable: accuracy rarely deviated from theoretical values by more than 20%, and precision rarely exceeded 33%. Determinations for biological samples were replicable irrespective of tissue (ovary or muscle). There was good linearity between analyte signal strength and theoretical concentration (median R2 ≥ 0.981 for all elements) across ranges typically encountered in environmental studies. Concentrations could not be consistently obtained (i.e., determined concentrations were typically below detection limits) for boron, vanadium, molybdenum, and cadmium in muscles, and arsenic in both ovaries and muscles; however, detection limits were sufficiently low for most environmental contexts. Further methodological refinement could include the incorporation of spiked standards to extend linear ranges, and fine-tuning instrument parameters to obtain smoother signal intensities for rare elements. The method presented promotes the use of low-impact sample collection methods while enabling high-quality determinations of elemental concentrations in biological tissues.

Thermodynamic, transport, and structural properties of hydrophobic deep eutectic solvents composed of tetraalkylammonium chloride and decanoic acid.

With the emergence of hydrophobic deep eutectic solvents (DESs), the scope of applications of DESs has been expanded to include situations in which miscibility with water is undesirable. Whereas most studies have focused on the applications of hydrophobic DESs from a practical standpoint, few theoretical works exist that investigate the structural and thermodynamic properties at the nanoscale. In this study, Molecular Dynamics (MD) simulations have been performed to model DESs composed of tetraalkylammonium chloride hydrogen bond acceptor and decanoic acid hydrogen bond donor (HBD) at a molar ratio of 1:2, with three different cation chain lengths (4, 7, and 8). After fine-tuning force field parameters, densities, viscosities, self-diffusivities, and ionic conductivities of the DESs were computed over a wide temperature range. The liquid structure was examined using radial distribution functions (RDFs) and hydrogen bond analysis. The MD simulations reproduced the experimental density and viscosity data from the literature reasonably well and were used to predict diffusivities and ionic conductivities, for which experimental data are scarce or unavailable. It was found that although an increase in the cation chain length considerably affected the density and transport properties of the DESs (i.e., yielding smaller densities and slower dynamics), no significant influence was observed on the RDFs and the hydrogen bonds. The self-diffusivities showed the following order for the mobility of the various components: HBD > anion > cation. Strong hydrogen bonds between the hydroxyl and carbonyl groups of decanoic acid and between the hydroxyl group of decanoic acid and chloride were observed to dominate the intermolecular interactions.

Transfer Reinforcement Learning Across Homotopy Classes

The ability for robots to transfer their learned knowledge to new tasks—where data is scarce—is a fundamental challenge for successful robot learning. While fine-tuning has been well-studied as a simple but effective transfer approach in the context of supervised learning, it is not as well-explored in the context of reinforcement learning. In this work, we study the problem of fine-tuning in transfer reinforcement learning when tasks are parameterized by their reward functions, which are known beforehand. We conjecture that fine-tuning drastically underperforms when source and target trajectories are part of different homotopy classes: We demonstrate that fine-tuning policy parameters across homotopy classes compared to fine-tuning within a homotopy class requires more interaction with the environment, and in certain cases is impossible. We propose a novel fine-tuning algorithm, Ease-In-Ease-Out fine-tuning, that consists of a relaxing stage and a curriculum learning stage to enable transfer learning across homotopy classes. Finally, we evaluate our approach on several robotics-inspired simulated environments and empirically verify that the Ease-In-Ease-Out fine-tuning method can successfully fine-tune in a sample-efficient way compared to existing baselines.

Venus upper atmosphere revealed by a GCM: II. Model validation with temperature and density measurements

An improved high resolution (96 longitude by 96 latitude points) ground-to-thermosphere version of the Institut Pierre-Simon Laplace (IPSL) Venus General Circulation Model (VGCM), including non-orographic gravity waves (GW) parameterization and fine-tuned non-LTE parameters, is presented here. We focus on the validation of the model built from a collection of data mostly from Venus Express (2006–2014) experiments and coordinated ground-based telescope campaigns, in the upper mesosphere/lower thermosphere of Venus (80–150 km). These simulations result in an overall better agreement with temperature observations above 90 km, compared with previous versions of the VGCM. Density of CO2 and light species, such as CO and O, are also comparable with observations in terms of trend and order of magnitude. Systematic biases in the temperature structure are found between 80 and 100 km approximately (e.g. GCM is 20 to 40 K warmer than measurements) and above 130 km at the terminator (e.g. GCM is up to 50 K colder than observed). Possible candidates for those discrepancies are the uncertainties on the collisional rate coefficients used in the non-LTE parameterization (above 130 km), and assumptions on the CO2 mixing ratio made for stellar/solar occultation retrievals. Diurnal and latitudinal distribution of dynamical tracers (i.e. CO and O) are also analyzed, in a region poorly constrained by wind measurements and characterized by high variability over daily to weekly timescale. Overall, our simulations indicate that a weak westward retrograde wind is present in the mesosphere, up to about 120 km, producing the CO bulge displacement toward 2 h–3 h in the morning, instead of piling up at the anti-solar point, as for an idealized sub-solar to anti-solar circulation. This retrograde imbalance is suggested to be produced by perturbations of a ~ 5 days Kelvin wave impacting the mesosphere up to 110 km (described in a companion paper Navarro et al., 2021), combined with GW westward acceleration in the lower thermosphere, mostly above 110 km. On the whole, these model developments point to the importance of the inclusion of the lower atmosphere, higher resolution and finely tuned parameterizations in GCM of the Venusian upper atmosphere, in order to shed light on existing observations.

PSpecteR: A User-Friendly and Interactive Application for Visualizing Top-Down and Bottom-Up Proteomics Data in R.

Visual examination of mass spectrometry data is necessary to assess data quality and to facilitate data exploration. Graphics provide the means to evaluate spectral properties, test alternative peptide/protein sequence matches, prepare annotated spectra for publication, and fine-tune parameters during wet lab procedures. Visual inspection of LC-MS data is constrained by proteomics visualization software designed for particular workflows or vendor-specific tools without open-source code. We built PSpecteR, an open-source and interactive R Shiny web application for visualization of LC-MS data, with support for several steps of proteomics data processing, including reading various mass spectrometry files, running open-source database search engines, labeling spectra with fragmentation patterns, testing post-translational modifications, plotting where identified fragments map to reference sequences, and visualizing algorithmic output and metadata. All figures, tables, and spectra are exportable within one easy-to-use graphical user interface. Our current software provides a flexible and modern R framework to support fast implementation of additional features. The open-source code is readily available (https://github.com/EMSL-Computing/PSpecteR), and a PSpecteR Docker container (https://hub.docker.com/r/emslcomputing/pspecter) is available for easy local installation.

4-Class MI-EEG Signal Generation and Recognition with CVAE-GAN

As the capability of an electroencephalogram’s (EEG) measurement of the real-time electrodynamics of the human brain is known to all, signal processing techniques, particularly deep learning, could either provide a novel solution for learning but also optimize robust representations from EEG signals. Considering the limited data collection and inadequate concentration of during subjects testing, it becomes essential to obtain sufficient training data and useful features with a potential end-user of a brain–computer interface (BCI) system. In this paper, we combined a conditional variational auto-encoder network (CVAE) with a generative adversarial network (GAN) for learning latent representations from EEG brain signals. By updating the fine-tuned parameter fed into the resulting generative model, we could synthetize the EEG signal under a specific category. We employed an encoder network to obtain the distributed samples of the EEG signal, and applied an adversarial learning mechanism to continuous optimization of the parameters of the generator, discriminator and classifier. The CVAE was adopted to adjust the synthetics more approximately to the real sample class. Finally, we demonstrated our approach take advantages of both statistic and feature matching to make the training process converge faster and more stable and address the problem of small-scale datasets in deep learning applications for motor imagery tasks through data augmentation. The augmented training datasets produced by our proposed CVAE-GAN method significantly enhance the performance of MI-EEG recognition.