Linear Assumption Research Articles

How much is enough? The role of R&D investment in the innovation process

AbstractWhile both production function (PF) and Absorptive Capacity (AC) explanations have played an important role in explaining the relationship between a firm's Research and Development (R&D) and its innovation, each has developed independently of the contributions of the other. The purpose of this study is to theoretically and empirically develop a concept of AC that incorporates the role of diminishing returns and external spillovers (i.e., strategic alliances) into a biotechnology firm's R&D‐innovation process. In using count estimations, this study finds that a firm's R&D‐innovation process is subject to a nonlinear ‐U‐shaped‐ learning process and that this process is moderated by its strategic partnerships. The contribution of this study is that it challenges the linearity assumptions and findings of AC research and that it offers a greater openness to PF explanations of the R&D‐innovation process.

Building entrepreneurial resilience during crisis using generative AI: An empirical study on SMEs

Recently, Gen AI has garnered significant attention across various sectors of society, particularly capturing the interest of small business due to its capacity to allow them to reassess their business models with minimal investment. To understand how small and medium-sized firms have utilised Gen AI-based tools to cope with the market's high level of turbulence caused by the COVID-19 pandemic, geopolitical crises, and economic slowdown, researchers have conducted an empirical study. Although Gen AI is receiving more attention, there remains a dearth of empirical studies that investigate how it influences the entrepreneurial orientation of firms and their ability to cultivate entrepreneurial resilience amidst market turbulence. Most of the literature offers anecdotal evidence. To address this research gap, the authors have grounded their theoretical model and research hypotheses in the contingent view of dynamic capability. They tested the research hypotheses using cross-sectional data from a pre-tested survey instrument, which yielded 87 useable responses from small and medium enterprises in France. The authors used variance-based structural equation modelling with the commercial WarpPLS 7.0 software to test the theoretical model. The study's findings suggest that Gen AI and EO have a significant influence on building entrepreneurial resilience as higher-order and lower-order dynamic capabilities. However, market turbulence has a negative moderating effect on the path that joins entrepreneurial orientation and entrepreneurial resilience. The results suggest that the assumption that high market turbulence will have positive effects on dynamic capabilities and competitive advantage is not always true, and the linear assumption does not hold, which is consistent with some scholars' assumptions. The study's results offer significant contributions to the contingent view of dynamic capabilities and open new research avenues that require further investigation into the non-linear relationship of market turbulence.

Lean Inventory Management and Performance of Milk Processors in Kenya

Dairy is a key contributor to the Kenyan economy and it helps in poverty alleviation and food nutrition in rural and urban areas. Dairy processing acts as a link between dairy farmers and the consumers in Kenya. However, milk processors have been facing numerous challenges that consequently affect their performance. Poor quality of milk, post-harvest losses resulting from inadequate cooling plants and seasonality of milk production in the country are among the problems the milk processors face. Raw milk, the main inventory for milk processors, is highly perishable and requires processing immediately after collection. However, this has not been the case. In fact, 54% of the invested capacities of milk processors remain underutilized. This calls for leanness in inventory that will ultimately reduce waste and improve firms’ performance through reduced complaints. This study was guided by resource orchestration and lean theory. The study was a census and the unit of analysis was all milk processors in Kenya licensed by the Kenya Dairy Board. A pilot study was conducted to enhance the questionnaire’s quality. A Cronbach Alpha of 0.8 was used. Before the regression was run, diagnostic tests were conducted to check for assumptions of linearity. An analysis of variance was conducted to test the variables’ significance and, ultimately, a multiple regression analysis was run. Maintain zero inventory and use of technology to order inventory and maintain zero inventory had positive and significant coefficients while collaboration with suppliers and members in a supply chain had a negative insignificant coefficient. The study concludes that zero inventory and use of technology to order have significant positive correlation coefficients. This implies that the two can help a firm reduce customer complaints and in return improve performance.

Predicting song popularity in the digital age through Spotifys data

This study delves into predicting song popularity on Spotify by analyzing a dataset of song features from 1986 to 2022. Using linear regression, this paper examines the influence of audio characteristics such as energy, danceability, speechiness, duration, and mode, alongside the year of release. The findings indicate that danceability, more recent release years, and longer track duration are positively associated with higher popularity levels. Conversely, songs in minor keys are more favored than those in major keys. These results highlight the significance of both intrinsic musical qualities and evolving listener preferences over time. The model's robustness is ensured through comprehensive diagnostic tests that validate the assumptions of linearity, normality, and homoscedasticity, confirming the predictive reliability of the identified factors. This research not only enhances the understanding of the dynamics driving music popularity but also provides valuable insights for artists and producers aiming to optimize their music for digital platforms. By focusing on the critical elements that resonate with contemporary audiences, stakeholders can better strategize their music releases to maximize listener engagement and success on streaming platforms.

On the Bruun Rule suitability for modelling shoreline change

Modelling and understanding shoreline response to sea-level rise over decadal to centennial timescales is crucial for managing sandy coasts. Globally, the Bruun Rule is one of the most common shoreline model applied to facilitate this. However, there are several limitations to the Bruun Rule, which can compromise coastal management decisions. Here, we provide a novel assessment of the suitability of the Bruun Rule for application in sandy coasts by applying it to hindcast and project shoreline change in three morphologically different sandy beaches over yearly and decadal timescales. We also compare the Bruun Rule predictions against associated observations, published predictions from a hybrid shoreline model, and historical shoreline changes. Results show that the Bruun Rule fails to accurately hindcast and provide realistic projections of future shoreline change in sandy coasts. We attribute this to the Bruun Rule's assumption of linearity between shoreline retreat and relative sea-level rise and its inability to account for nearshore circulation. We, therefore, recommend limiting the use of the Bruun Rule for informing policy and the management of sandy coasts.

Machine learning assisted Kriging to capture spatial variability in petrophysical property modelling

Machine learning (ML) models have been widely used for estimating formation geophysical properties. While ML models excel at capturing non-linear correlations among parameters, incorporating the impact of spatial variability into the training and testing process remains challenging. On the other hand, deterministic geostatistical models such as Kriging are capable of capturing spatial variability to a great extent but often rely on the assumption of linearity, making them less suitable for non-linear relations. In this study, a machine learning (Least Squared Support Vector Regression, LSSVR) model is integrated into Kriging interpolation to estimate the sonic compressional wave velocity with high spatial variability over a geological setting. This new methodology consists of two steps: (1) constructing an initial machine learning model using LSSVR and calculating the residuals, and (2) applying Kriging interpolation and gradually modifying the training data using Kriging estimation to optimize the machine learning model. The proposed model effectively captures the non-linear relation between auxiliary and target variables along with the spatial autocorrelation of data. To evaluate the performance of proposed hybrid model, a dataset comprising geophysical downhole logging measurements including density, gamma ray, depth and sonic velocity were extracted from 122 boreholes in a mine site located at the Bowen basin in Queensland, Australia. The data was averaged over a sandstone interval and divided into training and testing sets. Four models, namely linear regression, regression Kriging, LSSVR and the hybrid model were developed and compared with each other based on their R2 and RMSE values. The results demonstrate that the hybrid model outperforms the other models highlighting the effectiveness of employing a hybrid approach in capturing the spatial variability in petrophysical data over geological formations.

Mbtransfer: Microbiome intervention analysis using transfer functions and mirror statistics.

Time series studies of microbiome interventions provide valuable data about microbial ecosystem structure. Unfortunately, existing models of microbial community dynamics have limited temporal memory and expressivity, relying on Markov or linearity assumptions. To address this, we introduce a new class of models based on transfer functions. These models learn impulse responses, capturing the potentially delayed effects of environmental changes on the microbial community. This allows us to simulate trajectories under hypothetical interventions and select significantly perturbed taxa with False Discovery Rate guarantees. Through simulations, we show that our approach effectively reduces forecasting errors compared to strong baselines and accurately pinpoints taxa of interest. Our case studies highlight the interpretability of the resulting differential response trajectories. An R package, mbtransfer, and notebooks to replicate the simulation and case studies are provided.

Assessment of RANS-based turbulence models for isothermal confined swirling flow in a realistic can-type gas turbine combustor application

The present study assesses RANS-based turbulence models to simulate the isothermal confined swirling flow in a combustor representing a constituent can combustor of the can-annular configuration used in jet engines. The two-equation models (standard k−ϵ, realizable k−ϵ, standard k−ω, SST k−ω), and seven-equation model (linear pressure strain-Reynolds stress model, LPS-RSM), are assessed by comparing their predictions of mean axial velocity, mean transverse velocity, turbulent kinetic energy, and shear stress with the experimental data at two different positions (i.e., the primary and dilution hole planes) in the combustor. While the two-equation models generally have failed to predict the confined swirling flow at both positions accurately, the SST k−ω model yielded the most accurate, followed by standard k−ω and realizable k−ϵ models. The discrepancies between the computational and experimental results could be attributed to the isotropic turbulence assumptions, which, however, are invalid for confined swirling flows. Further, the two-equation model formulations cannot capture the intricacies of vortex flow and its interaction with the surroundings in confined swirling flows. The LPS-RSM, which considers turbulence anisotropy, showed some promise, although overpredicted results follow the trend with experimental values at the primary holes plane. However, at the dilution holes plane, the model overpredicted the velocity field, and underestimated the turbulence field, including turbulent kinetic energy and shear stress. These observed discrepancies can be ascribed to the pressure-strain correlation in the LPS-RSM, which assumes the pressure is a linear function of the strain-rate tensor. However, this linear assumption is quite simplistic for complex flows. Further, the influence of discretization (SOU and third-order MUSCL) schemes of convective terms is also assessed, and the differences in predictions resulted from MUSCL scheme having lower diffusion and superior ability to capture sharper gradients, however, did not translate into improving the solution accuracy. Hence, this study suggests that more advanced high-fidelity turbulence models (e.g., hybrid RANS-LES, LES, DNS) are needed to accurately predict the confined swirling flow in combustors.

Stability evaluation and design of negative stiffness amplifying damper-incorporated nonlinear structures

Negative stiffness-incorporated dampers have been widely accepted as effective vibration mitigation systems for flexible structural stiffness adjustment and enhanced energy dissipation. However, current investigations and optimal designs of negative stiffness-incorporated structures are limited to the linear elastic assumption of primary structures, potentially causing stability issues and unsatisfactory control performance. In this case, this study derives the stability criterion for negative stiffness amplifying damper (NSAD)-incorporated nonlinear structures and establishes stability-oriented design as well as modification formulae. The theoretical foundation of NSADs and nonlinear primary structures are initially introduced, based on which the stability criterion is derived according to the Routh–Hurwitz stability criterion. Following the existing design of linear elastic assumption, an endurance time acceleration series is generated subsequently to examine the proposed stability criterion for NSAD-equipped nonlinear structures. Consequently, the stability-oriented design method and corresponding modification formulae are proposed for near-field and far-field earthquakes, of which the effectiveness is validated through design cases. The results underscore the importance of considering structural nonlinearity for ensuring stability and designing NSAD-equipped nonlinear structures, demonstrating the feasibility of using the proposed correction factors to guarantee the stability of NSAD-equipped nonlinear structures subjected to multi-type earthquakes with various intensities. Additionally, the established design framework successively provides a high-efficiency energy dissipation device for seismic control of nonlinear structures, simultaneously generating easy-to-use design formulae for NSADs by combining the linear structure-based design and modification factors, facilitating direct design of nonlinear structures.

Exploring Multicomponent Phase Space to Discover New Materials

Multicomponent phase space has been shown to consist of an enormous number of materials with different compositions, the vast majority of which have never been made or investigated, with great potential, therefore, for the discovery of exciting new materials with valuable properties. At the same time, however, the enormous size of multicomponent phase space makes it far from straightforward to identify suitable strategies for exploring the plethora of potential material compositions and difficult, therefore, to be successful in discovering desirable new materials. Unfortunately, all our knowhow and understanding has been developed for materials with relatively few components in relatively limited proportions, with most of our scientific theories relying essentially on linear assumptions of component dilution and independence that no longer apply in concentrated multicomponent materials. Trial and error, controlled substitution, parameterisation, thermodynamic modelling, atomistic modelling and machine learning techniques have all been employed as methods of exploring multicomponent phase space, with varying levels of success, but ultimately none of these techniques has proved capable of delivering consistent or guaranteed results. This paper provides an overview of the different techniques that have been used to explore multicomponent phase space, indicates their main advantages and disadvantages, and describes some of their successes and failures.

Advances in Radar Signal Processing: Integrating Deep Learning Approaches

In the military, in daily life, and in scientific research, radar technology is widely used. Radar signal processing has long been an essential component in target detection and imaging. The development of deep learning technology in recent years has given radar signal processing new approaches and resources. With its exceptional feature extraction and pattern recognition skills, deep learning has made amazing strides and has been applied to radar signal processing to enhance tasks like target detection, tracking, and recognition. Traditional radar signal processing is based on models. It mainly uses the prior information of the model and related signal processing criteria to design signal processing methods. It uses Gaussian, linear, and stationary assumptions. The deep learning method is a data-based method that does not require prior knowledge of the model and can spontaneously find the relationship between the input of the algorithm and the expected output. This article will review traditional methods and deep learning methods in radar signal processing, focusing on the application and future development direction of deep learning methods in radar signal processing, briefly sorting out the research progress in recent years, and analyze some existing problems or shortcomings of existing methods.

On duration effects in non-life insurance pricing

The paper discusses duration effects on the consistency of mean parameter and dispersion parameter estimators in exponential dispersion families (EDFs) that are the standard models used for non-life insurance pricing. Focus is on the standard generalised linear model assumptions where both the mean and variance, conditional on duration, are linear functions in terms of duration. We derive simple convergence results that highlight consequences when the linear conditional moment assumptions are not satisfied. These results illustrate that: (i) the resulting mean estimators always have a relevant asymptotic interpretation in terms of the duration adjusted actuarially fair premium—a premium that only agrees with the standard actuarial premium using a duration equal to one, given that the expected value is linear in the duration; (ii) deviance based estimators of the dispersion parameter in an EDF should be avoided in favour of Pearson estimators; (iii) unless the linear moment assumptions are satisfied, consistency of dispersion and plug-in variance estimators can not be guaranteed and may result in spurious over-dispersion. The results provide explicit conditions on the underlying data generating process that will lead to spurious over-dispersion that can be used for model checking. This is illustrated based on real insurance data, where it is concluded that the linear moment assumptions are violated, which results in non-negligible spurious over-dispersion.

Regional divergent evolution of vegetation greenness and climatic drivers in the Sahel-Sudan-Guinea region: nonlinearity and explainable machine learning

IntroductionThe vegetation dynamics of the Sahel-Sudan-Guinea region in Africa, one of the largest transition zones between arid and humid zones, is of great significance for understanding regional ecosystem changes. However, a time-unvarying trend based on linear assumption challenges the overall understanding of vegetation greenness evolution and of tracking a complex ecosystem response to climate in the Sahel-Sudan-Guinea region.MethodsThis study first applied the ensemble empirical mode decomposition (EEMD) method to detect the time-varying trends in vegetation greenness based on normalized difference vegetation index (NDVI) data in the region during 2001–2020, and then identified the dominant climatic drivers of NDVI trends by employing explainable machine learning framework.ResultsThe study revealed an overall vegetation greening but a significant nonlinear spatio-temporal evolution characteristic over the region. Trend reversals, i.e., browning-to-greening and greening-to-browning, were dominant in approximately 60% of the study area. The browning-to-greening reversal was primarily observed in the southern Sahel, Congo Basin north of the Equator, and East Africa, with a breakpoint around 2008, while the greening-to-browning reversal was mainly observed in West Africa, with a breakpoint around 2011. The sustained greening primarily took place in northern Sahel, Central African Republic and South Sudan; while sustained browning clustered in central West Africa and Uganda, mainly in agricultural lands. Furthermore, the combination of Random Forest (RF) algorithm and the SHapley Additive exPlanations (SHAP) method could robustly model and reveal the relationships between the observed trends in NDVI and in climatic variables, also detected by applying EEMD. The results suggested that air temperature and precipitation were the most important climatic drivers controlling the NDVI trends across the Sahel-Sudan-Guinea region. The NDVI trends were more likely to have negative correlations with solar radiation and vapor pressure deficit in arid areas, while they could have positive correlations in humid areas. The study also found that large-scale climate changes induced by sea surface temperature (SST) anomalies had strong relationships with trend reversals in vegetation greenness at a sub-continental scale. These findings advanced the understanding of the impacts of climatic drivers on vegetation greenness evolution in the Sahel-Sudan-Guinea region.

Workforce thresholds and the non-linear association between registered nurse staffing and care quality in long-term residential care: A retrospective longitudinal study of English care homes with nursing

BackgroundCare needs amongst 425,000 dependent older residents in English care homes are becoming more complex. The quality of care in these homes is influenced by staffing levels, especially the presence of registered nurses (RNs). Existing research on this topic, often US-focused and relying on linear assumptions, has limitations. This study aims to investigate the non-linear relationship between RN staffing and care quality in English care homes using machine learning and administrative data from two major care home providers. MethodsA retrospective observational study was conducted using data from two English care home providers. Each was analysed separately due to variations in data reporting and care processes. Various care quality indicators and staffing metrics were collected for a 3.5-year period. Regression analysis and machine learning (random forest) were employed to identify non-linear relationships. Ethical approval was obtained for the study. ResultsUsing linear methods, higher skill mix – more care provided by RNs – was associated with lower incidence of adverse outcomes, such as urinary tract infections and hospitalisations. However, non-linear skill mix–outcome relationship modelling revealed both low and high skill mix levels were linked to higher risks. The effects of agency RN usage varied between providers, increasing risks in one but not the other. DiscussionThe study highlights the cost implications of increasing RN staffing establishments to improve care quality, suggesting a non-linear relationship and an optimal staffing threshold of around one-quarter of care provided by nurses. Alternative roles, such as care practitioners, merit exploration for meeting care demands whilst maintaining quality. This research underscores the need for a workforce plan for social care in England. It advocates for the incorporation of machine learning models alongside traditional regression-based methods. Our results may have limited generalisability to smaller providers and experimental research to redesign care processes effectively may be needed. ConclusionRNs are crucial for quality in care homes. Contrary to the assumption that higher nurse staffing necessarily leads to better care quality, this study reveals a nuanced, non-linear relationship between RN staffing and care quality in English care homes. It suggests that identifying an optimal staffing threshold, beyond which increasing nursing inputs may not significantly enhance care quality may necessitate reconsidering care system design and (human) resource allocation. Further experimental research is required to elucidate resource-specific thresholds and further strengthen evidence for care home staffing. Tweetable abstractHow much nursing care is needed to assure quality in care homes? Evidence from 2 English care home providers shows that nurse sensitive outcomes (an indicator of quality) are better when ~25 % of care is provided by nurses. Nurse shortages increase risks for residents.

Estimating and presenting hazard ratios and absolute risks from a Cox model with complex nonlinear interactions.

Interaction analysis is a critical component of clinical and public health research and represents a key topic in precision health and medicine. In applied settings, however, interaction assessment is usually limited to the test of a product term in a regression model and to the presentation of results stratified by levels of additional covariates. Stratification of results often relies on categorizing or making linearity assumptions for continuous covariates, with substantial loss of precision and of relevant information. In time-to-event analysis, moreover, interaction assessment is often limited to the multiplicative hazard scale by inclusion of a product term in a Cox regression model, disregarding the clinically relevant information that is captured by the absolute risk scale. In this paper we present a user-friendly procedure, based on the prediction of individual absolute risks from the Cox model, for the estimation and presentation of interactive effects on both the multiplicative and additive scales in survival analysis. We describe how to flexibly incorporate interactions with continuous covariates, which potentially operate in a nonlinear fashion, provide software for replicating our procedure, and discuss different approaches to deriving CIs. The presented approach will allow clinical and public health researchers to assess complex relationships between multiple covariates as they relate to a clinical endpoint, and to provide a more intuitive and precise depiction of the results in applied research papers focusing on interaction and effect stratification.

Efficiency enhancement techniques in finite element analysis: navigating complexity for agile design exploration

PurposeThis paper aims to comprehensively explore techniques for reducing solution time in finite element analysis (FEA), addressing the critical need for expediting computations to facilitate agile design exploration within project timelines.Design/methodology/approachDrawing from a wide array of literature sources, this paper synthesizes and analyzes various methodologies used to enhance the efficiency of FEA. Techniques are scrutinized in terms of their applicability, effectiveness and potential limitations.FindingsThe review signifies application of linear assumptions across multiple facets of analysis and delves into matrix order reduction strategies, geometry simplification, symmetry exploitation, submodeling and mesh attribute control. It reveals how these techniques can effectively reduce computational burdens while maintaining acceptable levels of accuracy.Research limitations/implicationsWhile this review provides a comprehensive overview of existing efficiency enhancement techniques in FEA, it acknowledges inherent limitations of any synthesis-based study. Future research should focus on refining these methodologies.Practical implicationsThe insights provided in this paper offer practical guidance for structural engineers and researchers seeking to optimize FEA workflows. By implementing these techniques, practitioners can expedite solution times and enhance their ability to explore design alternatives efficiently ultimately leading to cost savings and more robust structures.Originality/valueThis review contributes to the existing literature by offering a comprehensive synthesis of efficiency enhancement techniques in FEA. By highlighting the originality and value of each discussed methodology, this paper provides a roadmap for future research and practical implementation in the field of structural engineering.

Neural network survivability approach of a wave energy converter considering uncertainties in the prediction of future knowledge

To tune the wave energy converter (WEC) controller parameters such as damping to reduce the line force during extreme wave conditions, future knowledge of the line force is required. To achieve this, the incoming wave and system state should be predicted for a few seconds in the future. It is rather an arduous task to predict the future knowledge of waves and the system’s dynamic when dealing with breaking and steep waves, and the system is subject to various nonlinear forces. The classical model-based control strategies often rely on linear assumptions to estimate the WEC dynamics for the sake of simplicity. Unlike the model-based, the data-driven approaches are free from modeling errors and the algorithms are trained over the true and noisy data to predict non-linear system behaviors. Using data-driven approaches, we are able to model nonlinear dynamics. However, new questions emerge on the accuracy of the future wave and system state predictions, and how this uncertainty propagates into the final prediction of the line force. As incorrect damping may lead to excessive line force and detrimental damage to the system, these are the knowledge gaps that need to be addressed. The main purpose of this paper is to answer these questions through a survivability strategy for wave energy converters by providing a realistic perspective on the implementation of the neural network approaches by accounting for the errors in the input data. For this purpose, a series of neural networks is designed that first predicts the surface elevation for 0.36 s ahead, i.e. corresponding to 2 s in the full-scale WEC. This future knowledge of the wave elevation is then used to predict the system state (i.e. power take-off (PTO) translator position) for the same prediction horizon based on the PTO damping. This information is then fed to a convolutional neural network (CNN) that predicts the peak line force 0.36 s ahead. This paper also evaluates the predictive capabilities of neural network (NN) models, comparing them to classical autoregressive (AR) and Kalman filter methods. While the AR model exhibits slightly higher accuracy in fixed PTO system configurations, it falters in providing real-time predictions when system parameters undergo variations. In contrast, the NN prediction model excels by establishing a robust relationship between system configuration and output signals. Especially noteworthy is its ability to outperform classical methods with minimal training on a selective set of system configurations. Then, the sensitivity of the peak line force prediction to the uncertainties in the input data from NN models and the prediction horizon is analyzed. The neural network models are trained over the experimental data subjected to extreme sea states for a point absorber wave energy converter. The results suggest that the accuracy of the surface elevation prediction has an insignificant direct effect on the peak force prediction model. However, these uncertainties are reflected in the PTO translator position prediction, and the model is considerably sensitive to the accuracy of this prediction. This sensitivity nonetheless is less notable for higher PTO damping values.

The Impact of Entrepreneurial Orientation on Performance through Competence in Rattan Handicraft SMEs

This study aimed to analyze the relationship between entrepreneurial orientation to performance with competence as mediation. This quantitative study was conducted at Rattan Handicraft MSMEs in Gresik Regency, East Java, Indonesia. The sampling technique used census/full sampling with a sample of 82 MSME employees. The data analysis method uses Path analysis through several stages, including linearity assumption, measurement model test (outer model), structural model test (inner model), and hypothesis testing. The results of this study indicate an influence between entrepreneurial orientation and competence, the influence between entrepreneurial orientation on performance, and the influence between competence on performance, and competence can mediate the relationship between entrepreneurial orientation and performance. The practical contribution of this research is that it can Business development strategies as well as manpower development in MSMEs. The implications of this research can bring a deeper understanding of the dynamics involved in achieving performance excellence in the context of business and development strategies that can increase the success of MSMEs in achieving goals.

Lung function trajectories from school age to adulthood and their relationship with markers of cardiovascular disease risk

RationaleLung function in early adulthood is associated with subsequent adverse health outcomes.ObjectivesTo ascertain whether stable and reproducible lung function trajectories can be derived in different populations and investigate their association...

DESARROLLO DE MÉTODOS BASADOS EN REDES NEURONALES EN LA ESTIMACIÓN DE RECURSOS MINERALES

Due to economic and physical limitations, our understanding of mineral resources in a specific area of interest is limited and fragmented. Traditionally, this problem has been solved using the Kriging geostatistical method, where the ore grade is estimated at unmeasured locations using known values of the grade at surrounding points. The advantage of this method lies in the calculation of weights through a spatial variability model known as a variogram. However, the method is imperfect, as it is based on the assumption of stationarity, aditivity, linearity and potential subjectivity in variographic modelling. This study proposes to approach the mineral resource estimation problem as a regression problem using neural networks, which are not subject to the restrictions of stationarity, aditivity, linearity and spatial modelling of geostatistics methods. Kriging and a radial basis function neural network and a multilayer perceptron have been compared using different validation metrics. The results show that a properly trained neural network model, with appropriate labelling of the mineral grade and its input characteristics, achieves similar results to the geostatistical approach, with a significant reduction in time, while avoiding all the aforementioned assumptions. However, neural networks do not consider the spatial correlation of ore grade or reproduce it at the locations where it was measured, characteristics that have marked distrust in its industrial implementation and that are discussed in this article, finally proposing an adjustment between both approaches at a minimum sacrifice of time and labor costs. Keywords Neural networks; machine learning; geostatistics.