Development Of Data-driven Models Research Articles

A semi-supervised multi-task learning framework for cancer classification with weak annotation in whole-slide images.

Cancer region detection (CRD) and subtyping are two fundamental tasks in digital pathology image analysis. The development of data-driven models for CRD and subtyping on whole-slide images(WSIs) would mitigate the burden of pathologists and improve their accuracy in diagnosis. However, the existing models are facing two major limitations. Firstly, they typically require large-scale datasets with precise annotations, which contradicts with the original intention of reducing labor effort. Secondly, for the subtyping task, the non-cancerous regions are treated as the same as cancerous regions within a WSI, which confuses a subtyping model in its training process. To tackle the latter limitation, the previous research proposed to perform CRD first for ruling out the non-cancerous region, then train a subtyping model based on the remaining cancerous patches. However, separately training ignores the interaction of these two tasks, also leads to propagating the error of the CRD task to the subtyping task. To address these issues and concurrently improve the performance on both CRD and subtyping tasks, we propose a semi-supervised multi-task learning(MTL) framework for cancer classification. Our framework consists of a backbone feature extractor, two task-specific classifiers, and a weight control mechanism. The backbone feature extractor is shared by two task-specific classifiers, such that the interaction of CRD and subtyping tasks can be captured. The weight control mechanism preserves the sequential relationship of these two tasks and guarantees the error back-propagation from the subtyping task to the CRD task under the MTL framework. We train the overall framework in a semi-supervised setting, where datasets only involve small quantities of annotations produced by our minimal point-based(min-point) annotation strategy. Extensive experiments on four large datasets with different cancer types demonstrate the effectiveness of the proposed framework in both accuracy and generalization.

Italian, European, and international neuroinformatics efforts: An overview.

Neuroinformatics is a research field that focusses on software tools capable of identifying, analysing, modelling, organising and sharing multiscale neuroscience data. Neuroinformatics has exploded in the last two decades with the emergence of the Big Data phenomenon, characterised by the so-called 3Vs (volume, velocity and variety), which provided neuroscientists with an improved ability to acquire and process data faster and more cheaply thanks to technical improvements in clinical, genomic and radiological technologies. This situation has led to a 'data deluge', as neuroscientists can routinely collect more study data in a few days than they could in a year just a decade ago. To address this phenomenon, several neuroimaging-focussed neuroinformatics platforms have emerged, funded by national or transnational agencies, with the following goals: (i) development of tools for archiving and organising analytical data (XNAT, REDCap and LabKey); (ii) development of data-driven models evolving from reductionist approaches to multidimensional models (RIN, IVN, HBD, EuroPOND, E-DADS and GAAIN BRAIN); and (iii) development of e-infrastructures to provide sufficient computational power and storage resources (neuGRID, HBP-EBRAINS, LONI and CONP). Although the scenario is still fragmented, there are technological and economical attempts at both national and international levels to introduce high standards for open and Findable, Accessible, Interoperable and Reusable (FAIR) neuroscience worldwide.

Development of Data-Driven Models to Predict Biogas Production from Spent Mushroom Compost

In this study, two types of data-driven models were proposed to predict biogas production from anaerobic digestion of spent mushroom compost supplemented with wheat straw as a nutrient source. First, a k-nearest neighbours (k-NN) model (k = 1–10) was constructed. The optimal k value was determined using the cross-validation (CV) method. Second, a support vector machine (SVM) model was developed. The linear, quadratic, cubic, and Gaussian models were examined as kernel functions. The kernel scale was set to 6.93, while the box constraint (C) was optimized using the CV method. Results demonstrated that R2 for the k-NN model (k = 2) was 0.9830 at 35 °C and 0.9957 at 55 °C. The Gaussian-based SVM model (C = 1200) provided an R2 of 0.9973 at 35 °C and 0.9989 at 55 °C, which are slightly better than those achieved by k-NN. The Gaussian-based SVM model produced RMSE of 0.598 at 35 °C and 0.4183 at 55 °C, which are 58.4% and 49.5% smaller, respectively, than those produced by the k-NN. These findings imply that SVM modeling can be considered a robust technique in predicting biogas production from AD processes as they can be implemented without requiring prior knowledge of biogas production kinetics.

Artificial intelligence in renewable systems for transformation towards intelligent buildings

Carbon-neutrality transition in building sectors requires combinations of renewable systems and artificial intelligence (AI) for robustness, reliability, automation, and flexibility. In this study, a comprehensive review on AI applications in renewable systems, have been conducted, to report the current progress, tendency and challenges. Underlying learning mechanisms of AI-based various applications have been studied, in terms of modelling techniques on solar power forecasting, multi-level stochastic uncertainty analysis, smart controls, fault detection and diagnosis, single and multi-objective optimisations and intelligent buildings. Furthermore, AI techniques in renewable energy utilisations have been clearly reviewed, with respect to solar potential evaluation, multi-level stochastic uncertainty analysis, smart controls, fault detection and diagnosis, single and multi-objective optimisations. Results showed that, compared to physical models, the data-driven models show superiority in modelling simplification and modification, high prediction accuracy, and computational efficiency. In order to improve the system operational robustness under multi-level scenario uncertainty, a generic approach with artificial intelligence was proposed, including nonlinear data-driven model development, uncertainty-based energy performance prediction, stochastic uncertainty and statistical analysis. Furthermore, artificial intelligence can also be applied in renewable systems for security, reliability and stability. This study provides a clear roadmap on historical development, recent advances, cutting-edge techniques and progress, and future challenges of AI applications in renewable energy systems, paving path for developing smart and resilient renewable energy systems for decarbonization in buildings.

A gradient-based deep neural network model for simulating multiphase flow in porous media

Simulation of multiphase flow in porous media is crucial for the effective management of subsurface energy and environment-related activities. The numerical simulators used for modeling such processes rely on spatial and temporal discretization of the governing mass and energy balance partial-differential equations (PDEs) into algebraic systems via finite-difference/volume/element methods. These simulators usually require dedicated software development and maintenance, and suffer low efficiency from a runtime and memory standpoint for problems with multi-scale heterogeneity, coupled-physics processes or fluids with complex phase behavior. Therefore, developing cost-effective, data-driven models can become a practical choice, and in this work, we choose deep learning approaches as they can handle high dimensional data and accurately predict state variables with strong nonlinearity. In this paper, we describe a gradient-based deep neural network (GDNN) constrained by the physics related to multiphase flow in porous media. We tackle the nonlinearity of flow in porous media induced by rock heterogeneity, fluid properties, and fluid-rock interactions by decomposing the nonlinear PDEs into a dictionary of elementary differential operators. We use a combination of operators to handle rock spatial heterogeneity and fluid flow by advection. Since the augmented differential operators are inherently related to the physics of fluid flow, we treat them as first principles prior knowledge to regularize the GDNN training. We use the example of pressure management at geologic CO2 storage sites, where CO2 is injected in saline aquifers and brine is produced, and apply GDNN to construct a predictive model that is trained with physics-based simulation data and emulates the physics process. We demonstrate that GDNN can effectively predict the nonlinear patterns of subsurface responses, including the temporal and spatial evolution of the pressure and CO2 saturation plumes. We also successfully extend the GDNN to convolutional neural network (CNN), namely gradient-based CNN (GCNN), and validate its capability to improve the prediction accuracy. GDNN has great potential to tackle challenging problems that are governed by highly nonlinear physics and enable the development of data-driven models with higher fidelity.

A generalized machine learning workflow to visualize mechanical discontinuity

Accurate detection and mapping of mechanical discontinuity in materials has widespread industrial and research applications. We developed a generalized machine-learning framework for visualizing single mechanical discontinuity embedded in material of any composition, velocity, density, porosity, and size with limited data. The proposed visualization of discontinuity requires accurate estimations of the length, location, and orientation of the embedded discontinuity by processing multipoint wave-transmission measurements. k-Wave simulator is used to create a large dataset of elastic waveforms recorded during multi-point wave-transmission measurements through materials containing single mechanical discontinuity. k-Wave simulator considers the wave attenuation, dispersion, and mode conversion in wave motion. Discrete wavelet transform (DWT) and statistical feature extraction are essential for data preprocessing prior to the data-driven model development. DWT also minimizes the effect of noise. Using hyper-parameter tuning and cross validation, gradient boosting regression can visualize the mechanical discontinuity with an accuracy of 0.85, in terms of coefficient of determination. A double-layered neural network-based regression has better performance with an accuracy of 0.95. Use of convolutional neural network converts the predictive task from a waveform processing to an image processing problem. Convolutional neural network achieved a generalization performance of 0.91. The proposed generalized workflow requires robust simulation of wave propagation, signal processing, feature engineering, and model evaluation. Sensors closest to the source and those located opposite the source are the most significant for the desired visualization. Notably, the sensors closest to the source capture the non-linear associations, whereas the sensor on the border opposite to the source capture the linear associations between the measured waveforms and the properties of the mechanical discontinuity.

Patient-Specific Precision Injury Signatures to Optimize Orthopaedic Interventions in Multiply Injured Patients (PRECISE STUDY).

Optimal timing and procedure selection that define staged treatment strategies can affect outcomes dramatically and remain an area of major debate in the treatment of multiply injured orthopaedic trauma patients. Decisions regarding timing and choice of orthopaedic procedure(s) are currently based on the physiologic condition of the patient, resource availability, and the expected magnitude of the intervention. Surgical decision-making algorithms rarely rely on precision-type data that account for demographics, magnitude of injury, and the physiologic/immunologic response to injury on a patient-specific basis. This study is a multicenter prospective investigation that will work toward developing a precision medicine approach to managing multiply injured patients by incorporating patient-specific indices that quantify (1) mechanical tissue damage volume; (2) cumulative hypoperfusion; (3) immunologic response; and (4) demographics. These indices will formulate a precision injury signature, unique to each patient, which will be explored for correspondence to outcomes and response to surgical interventions. The impact of the timing and magnitude of initial and staged surgical interventions on patient-specific physiologic and immunologic responses will be evaluated and described. The primary goal of the study will be the development of data-driven models that will inform clinical decision-making tools that can be used to predict outcomes and guide intervention decisions.

Diversifying Tire-Defect Image Generation Based on Generative Adversarial Network

With the development of data-driven models, deep learning has been increasingly applied in the field of defect detection. However, the performance of deep learning models is greatly restricted by costly labeling and sample scarcity. One of the best approaches to solve the data imbalance problem is increasing the quantity and diversity of defect samples. Meanwhile, the current models based on generative adversarial network (GAN) cannot readily control the category and shape of generated defect samples, which results in inefficient data augmentation. Thus, to simultaneously achieve the category and shape adjustability of defect samples, a novel model named diversifying tire-defect GAN (DTD-GAN) is proposed in this article by integrating the advantages of latent space decomposition and feature decoupling algorithm. In this model, an auxiliary classifier is designed to control the defect category by discrete latent codes, and the mutual information maximization approach is adopted to diversify the tire-defect shape via continuous latent codes that greatly improve the diversity of the generated tire-defect samples. To illustrate the effectiveness of the proposed DTD-GAN model, extensive experiments on image quality and diversity assessment are carried out on the images generated by this model and other state-of-the-art approaches, and the achieved improvement in deep learning model performance in a tire-defect detection task is also discussed.

A data-centric approach to the study of system-level prognostics for cyber physical systems: application to safe UAV operations

Maintaining safe operations in cyber physical systems is a complex task that must account for system degradation over time, since unexpected failures can result in the loss of life and property. Operational failures may be attributed to component degradation and disturbances in the environment that adversely impact system performance. Components in a CPS typically degrade at different rates, and, therefore, require continual monitoring to avoid unexpected failures. Moreover, the effects of multiple degrading components on system performance may be hard to predict. Developing and maintaining accurate physics-based system models can be expensive. Typically, it is infeasible to run a true system to failure, so researchers and practitioners have resorted to using data-driven techniques to better evaluate the effect of degrading components on overall system performance. However, sufficiently organized datasets of system operation are not readily available; the output of existing simulations is not organized to facilitate the use of data-driven machine learning techniques for prognostics. As a step toward addressing this problem, in this paper, we develop a data management framework and an end-to-end simulation testbed to generate such data. The framework facilitates the development and comparison of various system-level prognostics algorithms. We adopt a standard data-centered design methodology, combined with a model based engineering approach, to create a data management framework that address data integrity problems and facilitates the generation of reproducible results. We present an ontological design methodology centered around assets, processes, and data, and, as a proof of concept, develop an unmanned aerial vehicle (UAV) system operations database that captures operational data for UAVs with multiple degrading components operating in uncertain environments. Aim: The purpose of this work is to provide a systematic approach to data generation, curation, and storage that supports studies in fault management and system-level prognostics for real-world and simulated operations. We use a data-driven simulation-based approach to enable reliable and reproducible studies in system-level prognostics. This is accomplished with a data management methodology that enforces constraints on data types and interfaces, and decouples various parts of the simulation to enable proper links with related metadata. The goal is to provide a framework that facilitates data analysis and the development of data-driven models for prognostics using machine learning methods. We discuss the importance of systematic data management framework to support data generation with a simulation environment that generates operational data. We describe a standard framework for data management in the context of run-to-failure simulations, and develop a database schema and an API in MATLAB® and Python to support system-level prognostics analyses. Methods: A systematic approach to defining a data management framework for the study of prognostics applications is a central piece of this work. A second important contribution is the design of a Monte Carlo simulation environment to generate run to failure data for CPS with multiple degrading components. We adopt a bottom-up approach, starting with requirements and specifications, then move into functionality and constraints. With this framework, we use a Monte-Carlo simulation approach to generate data for developing and testing a variety of system-level prognostics algorithms. Results: We have developed a data management framework that can handle high dimensional and complex data generated from real or simulated systems for the study of prognostics. In this paper, we show the advantages of a well-organized data management framework for tracking high-fidelity data with high confidence for complex, dynamic CPS. Such frameworks impose data logging discipline and facilitate downstream uses for developing and comparing different data-driven monitoring, diagnostics, and system-level prognostics algorithms. Conclusions: In this paper, we demonstrate the design, development, and use of an asset, process, and data management framework for the research to develop prognostics & health management applications. This work helps fill a gap for system-level remaining useful life studies by providing a comprehensive simulation environment that can generate run to failure data, and a data management architecture that addresses the needs for system-level prognostics research. The framework is demonstrated with a Monte-Carlo simulation of a UAV system that operates multiple flights under different environmental conditions and degradation sources. This architecture for data management will enable researchers to conduct more complex experiments for a variety of cyber physical systems applications.

Development of data-driven models for estimating the probability of high-concentration occurrence of naturally occurring radioactive materials in groundwater

High-concentration occurrence probability estimation models was developed to estimate 238U and 222Rn in groundwater using in-situ monitoring data (i.e., geological rock types, well depth, T, pH, Eh, EC, DO, and HCO3). The estimation models are based on a non-linear data-driven method to improve their effectiveness for applications in different estimation cases using various in-situ monitoring data. When developing the models, most sensitive in-situ monitoring data are selectively utilized to train the models, where various statistical and correlation analyses are applied to overcome challenges, including model overfitting during training, a highly nonlinear correlation between input and target variables, and poor training due to low-quality monitoring data. Based on statistical analysis results, all input variables, except for Eh and well depth, are used for developing the 238U and 222Rn estimation models, respectively. Actual data collected from groundwater quality monitoring networks of South Korea from 2007 to 2019 are used to validate the developed models. Although it is difficult to characterize 222Rn occurrence using the geochemical conditions of groundwater because the gaseous phase behavior of 222Rn highly depends on structural geology, the 222Rn estimation model achieves reasonable performance with more than 70% accuracy. In addition, compared to the 222Rn estimation model, the 238U estimation model achieves higher classification accuracy with approximately 80%. Consequently, we can confirm that the developed estimation models can be effectively used to estimate the probability of high-concentration risk of 238U and 222Rn in groundwater. Conclusively, the practical applicability of the developed models is wide, as the models have been developed using direct observable and real-time data.

Data-driven model development for large-eddy simulation of turbulence using gene-expression programing

We apply the gene-expression programing (GEP) method to develop subgrid-scale models for large-eddy simulations (LESs) of turbulence. The GEP model is trained based on Galilean invariants and tensor basis functions, and the training data are from direct numerical simulation (DNS) of incompressible isotropic turbulence. The model trained with GEP has been explicitly tested, showing that the GEP model can not only provide high correlation coefficients in a priori tests but also show great agreement with filtered DNS data when applied to LES. Compared to commonly used models like the dynamic Smagorinsky model and the dynamic mixed model, the GEP model provides significantly improved results on turbulence statistics and flow structures. Based on an analysis of the explicitly given model equation, the enhanced predictions are related to the fact that the GEP model is less dissipative and that it introduces high-order terms closely related to vorticity distribution. Furthermore, the GEP model with the explicit equation is straightforward to be applied in LESs, and its additional computational cost is substantially smaller than that of models trained with artificial neural networks with similar levels of predictive accuracies in a posteriori tests.

Active Learning Framework for Time-Series Classification of Vibration and Industrial Process Data

Data-based models built using machine learning solutions are becoming more prominent in the condition monitoring, maintenance, and prognostics fields. The capacity to build these models using a machine learning approach depends largely in the quality of the data. Of particular importance is the availability of labelled data, which describes the conditions that are intended to be identified. However, properly labelled data that is useful in many machine learning strategies is a scare resource. Furthermore, producing high-quality labelled data is expensive, time-consuming and a lot of times inaccurate given the uncertainty surrounding the labeling process and the annotators. 
 Active Learning (AL) has emerged as a semi-supervised approach that enables cost and time reductions of the labeling process. This approach has had a delayed adoption for time series classification given the difficulty to extract and present the time series information in such a way that it is easy to understand for the human annotator who incorporates the labels. This difficulty arises from the large dimensionality that many of these time series possess. This challenge is exacerbated by the cold-start problem, where the initial labelled dataset used in typical AL frameworks may not exist. Thus, the initial set of labels to be allocated to the time series samples is not available. This last challenge is particularly common on many condition monitoring applications where data samples of specific faults or problems does not exist.
 In this article, we present an AL framework to be used in the classification of time series from industrial process data, in particular vibration waveforms originated from condition monitoring applications. In this framework, we deal with the absence of labels to train an initial classification model by introducing a pre-clustering step. This step uses an unsupervised clustering algorithm to identify the number of labels and selects the points with a stronger group belonging as initial samples to be labelled in the active learning step. Furthermore, this framework presents two approaches to present the information to the annotator that can be via time-series imaging and automatic extraction of statistical features. Our work is motivated by the interest to facilitate the effort required for labeling time-series waveforms, while maintaining a high level of accuracy and consistency on those labels. In addition, we study the number of time-series samples that require to be labelled to achieve different levels of classification accuracy, as well as their confidence intervals. These experiments are carried out using vibration signals from a well-known rolling element bearing dataset and typical process data from a production plant. 
 An active learning framework that considers the conditions of the data commonly found in maintenance and condition monitoring applications while presenting the data in ways easy to interpret by human annotators can facilitate the generation reliable datasets. These datasets can, in turn, assist in the development of data-driven models that describe the many different processes that a machine undergoes.

A novel short-term load forecasting framework based on time-series clustering and early classification algorithm

With the development of data-driven models, extracting information from historical data for better energy forecasting is critically important for energy planning and optimization in buildings. Feature engineering is a key factor in improving the performance of forecasting models. Adding load pattern labels for different daily energy consumption patterns resulting from different time schedules and weather conditions can help improve forecasting accuracy. Traditionally, pattern labeling focuses mainly on finding a day similar to the forecasting day based on calendar or other information, such as weather conditions. The most intuitive approach for dividing historical time-series load into patterns is clustering; however, the pattern cannot be determined before the load is known. To address this problem, this study proposes a novel short-term load forecasting framework integrating an early classification algorithm that uses a stochastic algorithm to predetermine the load pattern of a forecasting day. In addition, a hybrid multistep method combining the strengths of single-step forecasting and recursive multistep forecasting is integrated into the framework. The proposed framework was validated through a case study using actual metered data. The results demonstrate that the early classification and proposed labeling strategy produce satisfactory forecasting accuracy and significantly improve the forecasting performance of the LightGBM model.

A machine learning-based methodology for multi-parametric solution of chemical processes operation optimization under uncertainty

Chemical process operation optimization aims at obtaining the optimal operating set-points by real-time solution of an optimization problem that embeds a steady-state model of the process. This task is challenged by unavoidable Uncertain Parameters (UPs) variations. MultiParametric Programming (MPP) is an approach for solving this challenge, where the optimal set-points must be updated online, reacting to sudden changes in the UPs. MPP provides algebraic functions describing the optimal solution as a function of the UPs, which allows alleviating large computational cost required for solving the optimization problem each time the UPs values vary. However, MPP applicability requires a well-constructed mathematical model of the process, which is not suited for process operation optimization, where complex, highly nonlinear and/or black-box models are usually used. To tackle this issue, this paper proposes a machine learning-based methodology for multiparametric solution of continuous optimization problems. The methodology relies on the offline development of data-driven models that accurately approximate the multiparametric behavior of the optimal solution over the UPs space. The models are developed using data generated by running the optimization using the original complex process model under different UPs values. The models are, then, used online to, quickly, predict the optimal solutions in response to UPs variation. The methodology is applied to benchmark examples and two case studies of process operation optimization. The results demonstrate the methodology effectiveness in terms of high prediction accuracy (less than 1% of NRMSE, in most cases), robustness to deal with problems of different natures (linear, bilinear, quadratic, nonlinear and/or black boxes) and significant reduction in the complexity of the solution procedure compared to traditional approaches (a minimum of 67% reduction in the optimization time).

Deep-freeze graph training for latent learning

Scientific and engineering advances are primarily driven by multi-tier conceptual constructs and conditional theoretical frameworks. The theories allow predictions of hypothetical system responses, given a set of approximate conditions (ranges of applicability) imposed on latent parameters that cannot be measured directly. Learning to estimate the latent variables (Latent Learning) helps to pinpoint the anticipated range-edge anomalies and improves the confidence in interpretation, interpolation and extrapolation of limited experimental data. Due to high dimensionality and extreme non-linearity of the materials science problems, very large datasets are typically required for conventional data-driven model development. The vital experimental data collection, particularly on microstructural phases, is very challenging, which makes it difficult to compile a high-quality database. Incorporation of the domain knowledge into the computational graph structure, initialization and optimization processes presents a viable mechanism for developing accurate models, with limited datasets. This study successfully utilized the approach to build the Deep Freeze Graph (DeepFreG) by mapping known causality relationships and by digitizing empirical domain knowledge for Latent Learning (LL), with specific applications in materials science.

Machine learning prediction models for battery-electric bus energy consumption in transit

Road gradient (g), initial state of charge (SoCi), drag coefficient (Cd), road condition (RC), HVAC, driver aggressiveness (Dagg.), passenger loading (PL), stop density (SD), average speed (Va), Multiple Regression Analysis (MLR), Radial Basis Function (RBF), Decision Tree Model (DT), Gradient Boosting Decision Tree (GBDT), Support Vector Machine (SVM), Multilayer Perception Neural Network (MLP-NN), and Radial Basis Neural Network (RBNN) • We developed and validated a simulation model to estimate the E C of BEBs. • A factorial experimental design generated BEB operation scenarios. • Seven data-driven models are developed to predict BEB energy consumption. • The models accommodate vehicular, operational, topological, and external parameters. • Multiple Regression and support vector machine models are deemed appropriate to predict BEB’s E C . The energy consumption (E C ) of battery-electric buses (BEB) varies significantly due to the intertwined relationships of vehicular, operational, topological, and external parameters. This variation is posing several challenges to predict BEB’s energy consumption. Several studies are calling for the development of data-driven models to address this challenge. This study develops and compares seven data-driven modelling techniques that cover both machine learning and statistical models. The models are based on a full-factorial experimental design ( n = 907,199 ) of a validated Simulink energy simulation model. The models are then used to predict E C using a testing dataset ( n = 169,344 ). The results show some minor discrepancies between the developed models. All models explained more than 90% of the energy consumption variance. Further, the results indicate that road gradient and the battery state of charge are the most influential factors on E C , while driver behaviour and drag coefficient have the lowest impact.

Biogas maximization using data-driven modelling with uncertainty analysis and genetic algorithm for municipal wastewater anaerobic digestion

Anaerobic digestion processes create biogases that can be useful sources of energy. The development of data-driven models of anaerobic digestion processes via operating parameters can lead to increased biogas production rates, resulting in greater energy production, through process modification and optimization. This study assessed processed and unprocessed input operating parameter variables for the development of regression models with transparent structures (‘white-box’ models) to: (1) estimate biogas production rates from municipal wastewater treatment plant (MWTP) anaerobic digestors; (2) compare their performances to artificial neural network (ANN) and adaptive network-based fuzzy inference system (ANFIS) models with opaque structures (‘black-box’ models) using Monte Carlo Simulation for uncertainty analysis; and (3) integrate the models with a genetic algorithm (GA) to optimize operating parameters for maximization of MWTP biogas production rates. The input variables were anaerobic digestion operating parameters from a MWTP including volatile fatty acids, total/fixed/volatile solids, pH, and inflow rate, which were processed via correlation tests and principal component analysis. Overall, the results indicated that the processed data did not improve regression model performances. Additionally, the developed non-linear regression model with the unprocessed inputs had the best performance based on values including R = 0.81, RMSE = 0.95, and IA = 0.89. However, this model was less accurate, but interestingly had less uncertainty, as compared to ANN and ANFIS models which indicates the compromise between model accuracy and uncertainty. Thus, all three models were coupled with GA optimization with maximum biogas production rate estimates of 22.0, 23.1, and 28.6 m3/min for ANN, ANFIS, and non-linear regression models, respectively.

Physics-informed deep learning for prediction of CO2 storage site response

Accurate prediction of the CO2 plume migration and pressure is imperative for safe operation and economic management of carbon storage projects. Numerical reservoir simulations of CO2 flow could be used for this purpose allowing the operators and stakeholders to calculate the site response considering different operational scenarios and uncertainties in geological characterization. However, the computational toll of these high-fidelity simulations has motivated the recent development of data-driven models. Such models are less costly, but may overfit the data and produce predictions inconsistent with the underlying physical laws. Here, we propose a physics-informed deep learning method that uses deep neural networks but also incorporates flow equations to predict a carbon storage site response to CO2 injection. A 3D synthetic dataset is used to show the effectiveness of this modeling approach. The model approximates the temporal and spatial evolution of pressure and CO2 saturation and predicts water production rate over time (outputs), given the initial porosity, permeability and injection rate (inputs). First, we establish a baseline using data-driven deep learning models namely, Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM). To build a physics-informed model, the loss term is modified using the constraints defined by a simplified form of the governing partial differential equations (conservation of mass coupled with Darcy's law for a two-phase flow system). Our results indicate that incorporating the domain knowledge significantly improves the accuracy of predictions. The proposed modeling approach can be integrated in CO2 storage management to accurately predict the critical site response indicators for a range of relevant input parameters, even when limited training data is available.

Data-driven model development to predict the aging of a Li-ion battery pack in electric vehicles representative conditions

An empirical generic Li-ion aging model, compatible with a large number of aging mechanisms without their a priori knowledge has been developed as well as a calibration methodology allowing its fast and automated parameter setting. This model has been applied to simulate the aging behavior of a 26 Ah cell. To train this model, a large aging test campaign has been conducted dedicated to both calibration and validation purposes. This one takes into account calendar, cycling, and their combinations. Based on the design of the aging campaign it is able to account for the effect of State Of Charge, temperature and current on aging. As its calibration is based on an automated process, it can be trained automatically and does not need expert knowledge for operation. Simulation data are validated to a 2% error in comparison to experimental data and is then validated for automotive applications.

Recent Advances in Imaging of Preclinical, Sporadic, and Autosomal Dominant Alzheimer's Disease.

Observing Alzheimer's disease (AD) pathological changes in vivo with neuroimaging provides invaluable opportunities to understand and predict the course of disease. Neuroimaging AD biomarkers also allow for real-time tracking of disease-modifying treatment in clinical trials. With recent neuroimaging advances, along with the burgeoning availability of longitudinal neuroimaging data and big-data harmonization approaches, a more comprehensive evaluation of the disease has shed light on the topographical staging and temporal sequencing of the disease. Multimodal imaging approaches have also promoted the development of data-driven models of AD-associated pathological propagation of tau proteinopathies. Studies of autosomal dominant, early sporadic, and late sporadic courses of the disease have shed unique insights into the AD pathological cascade, particularly with regard to genetic vulnerabilities and the identification of potential drug targets. Further, neuroimaging markers of b-amyloid, tau, and neurodegeneration have provided a powerful tool for validation of novel fluid cerebrospinal and plasma markers. This review highlights some of the latest advances in the field of human neuroimaging in AD across these topics, particularly with respect to positron emission tomography and structural and functional magnetic resonance imaging.