ML Solution Research Articles

Automated ASPECTS Segmentation and Scoring Tool: a Method Tailored for a Colombian TelestrokeNetwork.

To evaluate our two non-machine learning (non-ML)-based algorithmic approaches for detecting early ischemic infarcts on brain CT images of patients with acute ischemic stroke symptoms, tailored to our local population, to be incorporated in our telestroke software. One-hundred and thirteen acute stroke patients, excluding hemorrhagic, subacute, and chronic patients, with accessible brain CT images were divided into calibration and test sets. The gold standard was determined through consensus among three neuroradiologist. Four neuroradiologist independently reported Alberta Stroke Program Early CT Scores (ASPECTSs). ASPECTSs were also obtained using a commercial ML solution (CMLS), and our two methods, namely the Mean Hounsfield Unit (HU) relative difference (RELDIF) and the density distribution equivalence test (DDET), which used statistical analyze the of the HUs of each region and its contralateral side. Automated segmentation was perfect for cortical regions, while minimal adjustment was required for basal ganglia regions. For dichotomized-ASPECTSs (ASPECTS < 6) in the test set, the area under the receiver operating characteristic curve (AUC) was 0.85 for the DDET method, 0.84 for the RELDIF approach, 0.64 for the CMLS, and ranged from 0.71-0.89 for the neuroradiologist. The accuracy was 0.85 for the DDET method, 0.88 for the RELDIF approach, and was ranged from 0.83 - 0.96 for the neuroradiologist. Equivalence at a margin of 5% was documented among the DDET, RELDIF, and gold standard on mean ASPECTSs. Noninferiority tests of the AUC and accuracy of infarct detection revealed similarities between both DDET and RELDIF, and the CMLS, and with at least one neuroradiologist. The alignment of our methods with the evaluations of neuroradiologist and the CMLS indicates the potential of our methods to serve as supportive tools in clinical settings, facilitating prompt and accurate stroke diagnosis, especially in health care settings, such as Colombia, where neuroradiologist are limited.

Adapting UWB AoA estimation towards unseen environments using transfer learning and data augmentation

Ultra-wideband technology has become increasingly prevalent in localization systems, particularly with the emergence of multi-antenna systems capable of estimating both distance and angle of arrival (AoA) for incoming signals. However, most scientific research analyzes the accuracy of AoA in one specific environment for which the estimator is trained. In this paper, we analyze the performance of various AoA estimation algorithms, such as deep convolutional neural networks (DCNN), multiple signal classification (MUSIC), and phase difference of arrival (PDoA), in unseen environments. Prior work already demonstrated the superior performance of ML for AoA estimation compared to PDoA or MUSIC. We show that MUSIC, PDoA and ML solutions suffer from degradation in unseen environments at the 90th percentile of error, with ML-based AoA estimation degrading by about 14 degrees in unseen environments compared to 4 degrees for PDoA. We demonstrate that while PDoA more effectively corrects AoA at the median level in unseen environments, ML-based methods excel at correcting higher-percentile AoA errors, including outliers. Finally, we propose a novel framework to further improve the correction of AoA outliers for DCNN-based AoA estimators using data augmentation and transfer learning, resulting in a median angular error of only 5 degrees in unseen environments, even considering a field of view up to 90 degrees.

Predictive Models for Optimal Irrigation Scheduling and Water Management: A Review of AI and ML Approaches

Purpose: Maintaining agricultural output, protecting water supplies, and lessening environmental effects all depend on effective water management. Through a comprehensive review of the literature and an in-depth analysis of various AI and ML techniques, this paper aims to put light on the cutting-edge approaches used in irrigation scheduling predictive modeling. The goal of the research is to determine the advantages, disadvantages, and future directions of AI and ML-based irrigation management systems by means of a methodical analysis of various algorithms, data sources, and applications. Additionally, the study seeks to demonstrate how data-driven methods can enhance irrigation systems' sustainability, accuracy, and precision. Stakeholders in agriculture, water resource management, and environmental conservation can make well-informed decisions to maximize irrigation scheduling techniques by having a thorough understanding of the theoretical underpinnings and practical applications of predictive models. The study also attempts to tackle issues like scalability, model interpretability, and lack of data when implementing AI and ML solutions for practical irrigation management. In final form, this review's conclusions advance our understanding of how to use AI and ML to improve agricultural systems' resilience and water use efficiency, supporting adaptive and sustainable water management strategies in the face of rising water scarcity concerns and climate change. Design/Methodology/Approach: In order to gather information for this review study, several research articles from reliable sources were analyzed and compared. Objective: To provide the current research gaps in prediction models for the best irrigation scheduling and water management, and suggest using AI and ML techniques to fill in these gaps. Results/ Findings: In response to the growing challenges of water scarcity and climate change, the paper's findings highlight the transformative potential of AI and ML techniques in optimizing irrigation scheduling, enhancing agricultural resilience, increasing water use efficiency, and supporting adaptive and sustainable water management strategies. Originality/Value: This paper's uniqueness and significance come from its thorough analysis of AI and ML approaches in predictive modeling for ideal water management and irrigation scheduling. It also provides insights into new methods and their possible effects on resource optimization and agricultural sustainability. Type of Paper: Literature Review.

Improved financial forecasting via quantum machine learning

Quantum algorithms have the potential to enhance machine learning across a variety of domains and applications. In this work, we show how quantum machine learning can be used to improve financial forecasting. First, we use classical and quantum Determinantal Point Processes to enhance Random Forest models for churn prediction, improving precision by almost 6%. Second, we design quantum neural network architectures with orthogonal and compound layers for credit risk assessment, which match classical performance with significantly fewer parameters. Our results demonstrate that leveraging quantum ideas can effectively enhance the performance of machine learning, both today as quantum-inspired classical ML solutions, and even more in the future, with the advent of better quantum hardware.

A data-driven active learning approach to reusing ML solutions in scientific applications

Artificial intelligence can revolutionize scientific projects, but scientists face challenges in reusing, integrating, and deploying cost-effective and high-quality machine learning solutions. Determining suitable algorithms and parameters is difficult, especially for non-programmer scientists. Some algorithms, like deep learning-based methods, offer flexibility but require extensive training on annotated data. This poses a hurdle in labor-intensive tasks like biological image segmentation that relies on expert annotations.In this paper, we present a data-driven framework designed to assist scientists in selecting, reusing, and training machine learning solutions for microscopy image segmentation. The framework is based on establishing a mapping between object morphology features and the optimal segmentation algorithms and settings for individual objects. This mapping is iteratively refined through a combination of unsupervised learning and active learning iterations. To expedite convergence, objects are initially clustered based on their morphology. In each active learning iteration, the most informative and uncertain samples are selected and queried within a specific cluster. Through a biological case study, we demonstrate that our method enables the selection and training of segmentation algorithms specific to object types. Additionally, the selective requests for user input significantly reduce the number of user interactions required for this task.

Exploring the Conceptual Realm of Machine Learning in Small and Medium-sized Industries: A Qualitative Study

The pote­ntial of machine learning in small and medium busine­sses remains largely untappe­d. Through a qualitative study, it is e­xplored how the SMEs can practically apply ML to optimize processes and spark innovation. We­ aimed to demystify key conce­pts like supervised le­arning, data preparation, and model evaluation. Re­al-world examples across industries de­monstrate ML's versatility, from enhancing de­cision-making to improving efficiency. Our rese­arch highlights the transformative power of ML, e­specially for resource-constraine­d SMEs. With some guidance, eve­n small teams can implement ML solutions that unlock ne­w opportunities. Though adoption has barriers, from data to skills, ML’s value for SMEs is cle­ar. With a strategic approach, companies of all sizes can tap into its possibilitie­s. Current re­search on implementing machine­ learning in small and medium ente­rprises has gaps. More investigation is re­quired to understand adoption challenge­s fully, highlight successes, and mee­t unique needs. A qualitative­ approach that explores expe­riences and perspe­ctives can provide those rich insights. Some­ small and medium companies use machine­ learning, but many face hurdles adopting it. Re­search case studies showcase­ machine learning success storie­s, though each company's path differs. By understanding individual difficultie­s, researchers can he­lp more small and medium enterprises use machine learning appropriate­ly. This literature revie­w examines machine le­arning models, adoption trends, triumphs, and example­s in small and medium enterprise­s. Moreove­r, it also examines the advantage­s and obstacles SMEs encounter whe­n adopting ML tactics.

Neural Network‐based Method to Compute Recommended Cruise Level and Optimal Climb Location

AbstractSystem engineering towards utilizing machine learning solutions in the aviation domain is a fast‐progressing discipline. This paper tries to bring out the system engineering and implementation aspects of ML solution as an improvement in a sub‐system of an existing aviation system. The flight management systems (FMSs) compute the recommended cruise level (RCL) considering the aircraft's gross weight, its speed schedule, and the forecast wind. The RCL increases with distance in the no wind conditions as the fuel get consumed, reducing the gross aircraft weight. The RCL profile may have frequent climbs and descents in the presence of winds, which is not desired for flight operations.The document proposes a neural network‐based method for computing an operationally acceptable cruise profile considering the wind and temperature weather forecast. The method recommends step climbs and step descents (optimal level and optimal step location) in such a way that they are separated by a large enough cruise phase. From system engineering point of view two guiding documents for such an implementation to look forward are “EASA Concept Paper: First usable guidance for Level 1 machine learning applications” [12] and “Process Standard for Development and Certification/Approval of Aeronautical Safety‐Related Products Implementing AI” [13].Two neural networks have been created to estimate wind compensated ECON cruise speeds and required time‐fuel to cover the next 500NM distance. The RCL profiles are generated for various test wind scenarios for Airbus A220 aircraft. It has been recorded the proposed neural network‐based method is better at recommending a cruise altitude and it is 9X faster than the existing method to calculate the RCL and optimal climb location. The neural network‐based design also helped in reducing the memory footprint by more than 100 times as there is no requirement of storing the performance database files required by the existing system. Only the parameters of the trained neural networks are required to be stored. Most importantly, the RCL computed by the proposed neural network‐based method yields a smaller operational cost than that determined by the existing method. The proposed method is designed to replace the existing traditional subsystem to give an overall improved performance while the functionalities of other subsystems remain unaffected.

Refinements on the exact method to solve the numerical difficulties in fitting the log binomial regression model for estimating relative risk

Fitting a log binomial regression model using standard software can result in numerical difficulties because its functional form does not restrict the estimated probabilities to values not exceeding unity. The common approaches to resolve the issue are to introduce a constraint to limit the results of each iteration. However, if the ML solution lies on the boundary of the allowable parameter space, some fitted probabilities are equal to unity (named as boundary vector). The common approaches have trouble dealing with those boundary vectors and can only reach somewhere before the ML solution. Previously a remedy has been proposed, but without the details necessary to implement it. Here we provide these details, including formulae for estimating the covariances essential to implement the method, an explanation of inter-dependency between coefficient estimates, and a proof that the method can be implemented in general. Code written for R implements choice of fitting algorithms, finding appropriate starting values, identifying the covariate vector(s) with a fitted probability of unity, and strategies for covariate ordering to handle the issues caused by the common values in two distinct boundary vectors. The model-fitting results are compared with two alternative methods by simulation and example data.

Designing Smart Antennas Using Machine Learning Algorithms

Smart antenna technologies improve spectral efficiency, security, energy efficiency, and overall service quality in cellular networks by utilizing signal processing algorithms that provide radiation beams to users while producing nulls for interferers. In this paper, the performance of such ML solutions as the support vector machine (SVM) algorithm, the artificial neural network (ANN), the ensemble algorithm (EA), and the decision tree (DT) algorithm used for forming the beam of smart antennas are compared. A smart antenna array made up of 10 half-wave dipoles is considered. The ANN method is better than the remaining approaches when it comes to achieving beam and null directions, whereas EA offers better performance in terms of reducing the side lobe level (SLL). The maximum SLL is achieved using EA for all the user directions. The performance of the ANN algorithm in terms of forming the beam of a smart antenna is also compared with that of the variable-step size adaptive algorithm.

OpenData: A Framework to Train and Deploy ML Solutions in Wide-Area Networks

OpenData: A Framework to Train and Deploy ML Solutions in Wide-Area Networks

A machine learning approach for player and position adjusted expected goals in football (soccer)

Football is a very result-driven industry, with goals being rarer than in most sports, so having further parameters to judge the performance of teams and individuals is key. Expected Goals (xG) allow further insight than just a scoreline. To tackle the need for further analysis in football, this paper uses machine learning methods that are developed and applied to Football Event data. The proposed solution utilises StatsBomb as the data provider and an industry benchmark to tune the models in the right direction. The proposed ML solution for xG is further used to tackle the age-old cliche of: ‘the ball has fallen to the wrong guy there’. To investigate this, we tackle Positional Adjusted xG, splitting the training data into Forward, Midfield, and Defence to provide insight into player qualities based on their positional sub-group. Positional Adjusted xG successfully predicts and proves that more attacking players are better at accumulating xG. Finally, this study has further developments into Player Adjusted xG to prove that Lionel Messi is statistically at a higher efficiency level than the average footballer. Thanks to this analysis, we conclude that the Messi model performs 347 xG higher than the general model outcome.

Cyberbullying detection for low-resource languages and dialects: Review of the state of the art

The struggle of social media platforms to moderate content in a timely manner, encourages users to abuse such platforms to spread vulgar or abusive language, which, when performed repeatedly becomes cyberbullying — a social problem taking place in virtual environments, yet with real-world consequences, such as depression, withdrawal, or even suicide attempts of its victims. Systems for the automatic detection and mitigation of cyberbullying have been developed but, unfortunately, the vast majority of them are for the English language, with only a handful available for low-resource languages. To estimate the present state of research and recognize the needs for further development, in this paper we present a comprehensive systematic survey of studies done so far for automatic cyberbullying detection in low-resource languages. We analyzed all studies on this topic that were available.We investigated more than seventy published studies on automatic detection of cyberbullying or related language in low-resource languages and dialects that were published between around 2017 and January 2023. There are 23 low-resource languages and dialects covered by this paper, including Bangla, Hindi, Dravidian languages and others. In the survey, we identify some of the research gaps of previous studies, which include the lack of reliable definitions of cyberbullying and its relevant subcategories, biases in the acquisition, and annotation of data. Based on recognizing those research gaps, we provide some suggestions for improving the general research conduct in cyberbullying detection, with a primary focus on low-resource languages. Based on those proposed suggestions, we collect and release a cyberbullying dataset in the Chittagonian dialect of Bangla and propose a number of initial ML solutions trained on that dataset. In addition, pre-trained transformer-based the BanglaBERT model was also attempted. We conclude with additional discussions on ethical issues regarding such studies, highlight how our survey improves on similar surveys done in the past, and discuss the usefulness of recently popular AI-enhanced tools for streamlining such scientific surveys.

FPIC: A Novel Semantic Dataset for Optical PCB Assurance

Outsourced PCB fabrication necessitates increased hardware assurance capabilities. Several assurance techniques based on AOI have been proposed that leverage PCB images acquired using digital cameras. We review state-of-the-art AOI techniques and observe a strong, rapid trend toward ML solutions. These require significant amounts of labeled ground truth data, which is lacking in the publicly available PCB data space. We contribute the FPIC dataset to address this need. Additionally, we outline new hardware security methodologies enabled by our dataset.

Controlling Media Player Using Hand Gestures

Computer usage is increasing rapidly day by day but the input devices are limited and to access them, we need to be near the screen. To overcome this problem and control the screen, we can use hand gestures. For every operation, we used different hand gestures. We proposed a python program to control the media player through hand gestures. In this method, we used libraries like OpenCV, Media Pipe, PyAuto GUI, and other libraries to capture the video, provide ready-to-use ML solutions and automate your GUI and programmatically control your keyboard and mouse. Hand gestures will be used as the input for providing natural interaction by reducing external hardware interaction. The whole process is divided into two steps. Firstly, gesture recognition through the camera is done by OpenCV and media Pipe helps to identify the gesture b its position, and the respective command is executed. Secondly, PyAuto GUI is used to automate the keyboard and controls the media player.

Controlling Media Player Using Hand Gestures

Computer usage is increasing rapidly day by day but the input devices are limited and to access them, we need to be near the screen. To overcome this problem and control the screen, we can use hand gestures. For every operation, we used different hand gestures. We proposed a python program to control the media player through hand gestures. In this method, we used libraries like OpenCV, Media Pipe, PyAuto GUI, and other libraries to capture the video, provide ready-to-use ML solutions and automate your GUI and programmatically control your keyboard and mouse. Hand gestures will be used as the input for providing natural interaction by reducing external hardware interaction. The whole process is divided into two steps. Firstly, gesture recognition through the camera is done by OpenCV and media Pipe helps to identify the gesture b its position, and the respective command is executed. Secondly, PyAuto GUI is used to automate the keyboard and controls the media player.

Intelligent Healthcare Management: Advancing Healthcare with Integrated AI and ML Solutions

The proposed system, leveraging the power of big data, telecommunication technologies, and wearable sensors, presents a unique opportunity to transform the healthcare industry. It fosters a seamless connection between patients, wearable sensors, caregivers, and providers through the innovative use of Information and Communication Technology (ICT) and software. This is of utmost importance in developing countries, where the healthcare sector grapples with economic challenges amplified by a burgeoning population and a surging demand for quality care, particularly for the elderly. The urgency for new technologies and tools to bolster the healthcare of senior citizens has reached a tipping point. Breakthroughs in wireless technology, miniaturization, computing power, and processing have paved the way for the development of interconnected medical devices. This proposal introduces a cutting-edge healthcare monitoring system for tracking the activities of elderly individuals, harnessing the power of the Hadoop MapReduce technique for the parallel processing of large datasets. Data from wearable sensors placed on the subjects' left ankle, right arm, and chest are transmitted to a cloud platform and a data analytics layer via Internet of Medical Things (IoMT) devices. The data is then divided into small chunks and processed as Map tasks.

Enabling real time big data solutions for manufacturing at scale

Today we create and collect more data than we have in the past. All this data comes from different sources, including social media platforms, our phones and computers, healthcare gadgets and wearable technology, scientific instruments, financial institutions, the manufacturing industry, news channels, and more. When these data are analyzed in a real-time nature, it offers businesses the opportunity to take quick action in business-development processes (B2B, B2C), gain a different perspective, and better understand applications, creating new opportunities. While changing their sales and marketing strategies, businesses are now able to manage the data they collect in real-time to transform themselves, to record them in a healthy way, to analyze and evaluate data-based processes, and to determine their digital transformation roadmaps, their interactions with their customers, sectoral diffraction, application, and analysis. They want to accelerate the transformation processes within the technology triangle. Thus, big data, recently called as small and wide data, is at the center of everything and becomes an important application for digital transformation. Digital transformation helps companies embrace change and stay competitive in an increasingly digital world. The value of big data in manufacturing, independent from sectoral variations, comes from its ability to combine both in an organization’s efforts to both digitize and automate its end-to-end business operations. In this study, the current digitalization and automation applications of one of the plastic injection-based manufacturing companies at scale will be discussed. Presented open-source-based big data analytics platform, DataCone, that increases data processing efficiency, storage optimization, encourages innovation for real time monitorization and analytics, and support new business models in different industry segments will be demonstrated and discussed. Thus, development and applied ML solutions will be discussed providing important prospects for the future.

Correction to: Review of ML and AutoML Solutions to Forecast Time-Series Data

Correction to: Review of ML and AutoML Solutions to Forecast Time-Series Data

On the Support Recovery of Jointly Sparse Gaussian Sources via Sparse Bayesian Learning

In this work, we provide non-asymptotic, probabilistic guarantees for successful recovery of the common nonzero support of jointly sparse Gaussian sources in the multiple measurement vector (MMV) problem. The support recovery problem is formulated as the marginalized maximum likelihood (or type-II ML) estimation of the variance hyperparameters of a joint sparsity inducing Gaussian prior on the source signals. We derive conditions under which the resulting nonconvex constrained optimization perfectly recovers the nonzero support of a joint-sparse Gaussian source ensemble with arbitrarily high probability. The support error probability decays exponentially with the number of MMVs at a rate that depends on the smallest restricted singular value and the nonnegative null space property of the self Khatri-Rao product of the sensing matrix. Our analysis confirms that nonzero supports of size as high as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$O(m^{2})$ </tex-math></inline-formula> are recoverable from <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$m$ </tex-math></inline-formula> measurements per sparse vector. Our derived sufficient conditions for support consistency of the proposed constrained type-II ML solution also guarantee the support consistency of any global solution of the multiple sparse Bayesian learning (M-SBL) optimization whose nonzero coefficients lie inside a bounded interval. For the case of noiseless measurements, we further show that a single MMV is sufficient for perfect recovery of the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k$ </tex-math></inline-formula> -sparse support by M-SBL, provided all subsets of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$k + 1$ </tex-math></inline-formula> columns of the sensing matrix are linearly independent.

Review of ML and AutoML Solutions to Forecast Time-Series Data.

Time-series forecasting is a significant discipline of data modeling where past observations of the same variable are analyzed to predict the future values of the time series. Its prominence lies in different use cases where it is required, including economic, weather, stock price, business development, and other use cases. In this work, a review was conducted on the methods of analyzing time series starting from the traditional linear modeling techniques until the automated machine learning (AutoML) frameworks, including deep learning models. The objective of this review article is to support identifying the time-series forecasting challenge and the different techniques to meet the challenge. This work can be additionally an assist and a reference for researchers and industries demanding to use AutoML to solve the problem of forecasting. It identifies the gaps of the previous works and techniques used to solve the problem of forecasting time series.