ML Applications Research Articles

Artificial Intelligence for Epigenetics: Towards Personalized Medicine.

Epigenetics is a field of biological sciences focused on the study of reversible, heritable changes in gene function, not due to modifications of the genomic sequence. These changes are the result of a complex cross-talk between several molecular mechanisms that is in turn orchestrated by genetic and environmental factors. The epigenetic profile captures the unique regulatory landscape and the exposure to environmental stimuli of an individual. It thus constitutes a valuable reservoir of information for personalized medicine, which is aimed at customizing health-care interventions based on the unique characteristics of each individual. Nowadays, the complex milieu of epigenomic marks can be studied at the genome-wide level thanks to massive, high-throughput technologies. This new experimental approach is opening up new and interesting knowledge perspectives. However, the analysis of these complex omic data requires to face important analytic issues. Artificial Intelligence, and in particular Machine Learning, are emerging as powerful resources to decipher epigenomic data. In this review, we will first describe the most used ML approaches in epigenomics. We then will recapitulate some of the recent applications of ML to epigenomic analysis. Finally, we will provide some examples of how the ML approach to epigenetic data can be useful for personalized medicine.

Application of Machine Learning to Electroencephalography for the Diagnosis of Primary Progressive Aphasia: A Pilot Study.

Background. Primary progressive aphasia (PPA) is a neurodegenerative syndrome in which diagnosis is usually challenging. Biomarkers are needed for diagnosis and monitoring. In this study, we aimed to evaluate Electroencephalography (EEG) as a biomarker for the diagnosis of PPA. Methods. We conducted a cross-sectional study with 40 PPA patients categorized as non-fluent, semantic, and logopenic variants, and 20 controls. Resting-state EEG with 32 channels was acquired and preprocessed using several procedures (quantitative EEG, wavelet transformation, autoencoders, and graph theory analysis). Seven machine learning algorithms were evaluated (Decision Tree, Elastic Net, Support Vector Machines, Random Forest, K-Nearest Neighbors, Gaussian Naive Bayes, and Multinomial Naive Bayes). Results. Diagnostic capacity to distinguish between PPA and controls was high (accuracy 75%, F1-score 83% for kNN algorithm). The most important features in the classification were derived from network analysis based on graph theory. Conversely, discrimination between PPA variants was lower (Accuracy 58% and F1-score 60% for kNN). Conclusions. The application of ML to resting-state EEG may have a role in the diagnosis of PPA, especially in the differentiation from controls. Future studies with high-density EEG should explore the capacity to distinguish between PPA variants.

FLASH: F ast Neura l A rchitecture S earch with H ardware Optimization

Neural architecture search (NAS) is a promising technique to design efficient and high-performance deep neural networks (DNNs). As the performance requirements of ML applications grow continuously, the hardware accelerators start playing a central role in DNN design. This trend makes NAS even more complicated and time-consuming for most real applications. This paper proposes FLASH, a very fast NAS methodology that co-optimizes the DNN accuracy and performance on a real hardware platform. As the main theoretical contribution, we first propose the NN-Degree, an analytical metric to quantify the topological characteristics of DNNs with skip connections (e.g., DenseNets, ResNets, Wide-ResNets, and MobileNets). The newly proposed NN-Degree allows us to do training-free NAS within one second and build an accuracy predictor by training as few as 25 samples out of a vast search space with more than 63 billion configurations. Second, by performing inference on the target hardware, we fine-tune and validate our analytical models to estimate the latency, area, and energy consumption of various DNN architectures while executing standard ML datasets. Third, we construct a hierarchical algorithm based on simplicial homology global optimization (SHGO) to optimize the model-architecture co-design process, while considering the area, latency, and energy consumption of the target hardware. We demonstrate that, compared to the state-of-the-art NAS approaches, our proposed hierarchical SHGO-based algorithm enables more than four orders of magnitude speedup (specifically, the execution time of the proposed algorithm is about 0.1 seconds). Finally, our experimental evaluations show that FLASH is easily transferable to different hardware architectures, thus enabling us to do NAS on a Raspberry Pi-3B processor in less than 3 seconds.

Applied Machine Learning for Developing Next‐Generation Functional Materials

AbstractMachine learning (ML) is a versatile technique to rapidly and efficiently generate insights from multidimensional data. It offers a much‐needed avenue to accelerate the exploration and investigation of new materials to address time‐sensitive global challenges such as climate change. The availability of large datasets in recent years has enabled the development of ML algorithms for various applications including experimental/device optimization and material discovery. This perspective provides a summary of the recent applications of ML in material discovery in a range of fields, from optoelectronics to batteries and electrocatalysis, as well as an overview of the methods behind these advances. The paper also attempts to summarize some key challenges and trends in current research methodologies.

Machine Learning: An Advanced Platform for Materials Development and State Prediction in Lithium‐Ion Batteries

Lithium-ion batteries (LIBs) are vital energy-storage devices in modern society. However, the performance and cost are still not satisfactory in terms of energy density, power density, cycle life, safety, etc. To further improve the performance of batteries, traditional "trial-and-error" processes require a vast number of tedious experiments. Computational chemistry and artificial intelligence (AI) can significantly accelerate the research and development of novel battery systems. Herein, a heterogeneous category of AI technology for predicting and discovering battery materials and estimating the state of the battery system is reviewed. Successful examples, the challenges of deploying AI in real-world scenarios, and an integrated framework are analyzed and outlined. The state-of-the-art research about the applications of ML in the property prediction and battery discovery, including electrolyte and electrode materials, are further summarized. Meanwhile, the prediction of battery states is also provided. Finally, various existing challenges and the framework to tackle the challenges on the further development of machine learning for rechargeable LIBs are proposed.

Designing production-friendly machine learning

Building production ML applications is difficult because of their resource cost and complex failure modes. I will discuss these challenges from two perspectives: the Stanford DAWN Lab and experience with large-scale commercial ML users at Databricks. I will then present two emerging ideas to help address these challenges. The first is "ML platforms", an emerging class of software systems that standardize the interfaces used in ML applications to make them easier to build and maintain. I will give a few examples, including the open-source MLflow system from Databricks [3]. The second idea is models that are more "production-friendly" by design. As a concrete example, I will discuss retrieval-based NLP models such as Stanford's ColBERT [1, 2] that query documents from an updateable corpus to perform tasks such as question-answering, which gives multiple practical advantages, including low computational cost, high interpretability, and very fast updates to the model's "knowledge". These models are an exciting alternative to large language models such as GPT-3.

Machine learning on sustainable energy: A review and outlook on renewable energy systems, catalysis, smart grid and energy storage

This study presents a broad view of the current state of the art of ML applications in the manufacturing sectors that have a considerable impact on sustainability and the environment, namely renewable energies (solar, wind, hydropower, and biomass), smart grids, the industry of catalysis and power storage and distribution. Artificial neural networks are the most preferred techniques over other ML algorithms because of their generalization capabilities. Demands for ML techniques in the energy sectors will increase considerably in the coming years, since there is a growing demand of academic programmes related to artificial intelligence in science, math, and engineering. Data generation, management, and safety are expected to play a key role for the successful implementation of ML algorithms that can be shared by major stakeholders in the energy sector, thereby promoting the development of ambitious energy management projects. New algorithms for producing reliable data and the addition of other sources of information (e.g., novel sensors) will enhance flow of information between ML and systems. It is expected that unsupervised and reinforcement learning will take a central role in the energy sector, but this will depend on the expansion of other major fields in data science such as big data analytics. Massive implementations, specialized algorithms, and new technologies like 5G will promote the development of sustainable applications of ML in non-industrial applications for energy management.

OTHR-15. Assessment of TRIPOD adherence in articles developing machine learning models for differentiation of glioma from brain metastasis

PurposeMachine learning (ML) applications in predictive models in neuro-oncology have become an increasingly investigated subject of research. For their incorporation into clinical practice, rigorous assessment is needed to reduce bias. Several reports have indicated utility of ML applications in differentiation of glioma from brain metastasis. However, a systematic assessment of quality of methodology and reporting in these studies has not been done yet. We examined the adherence of 29 published reports in this field to the TRIPOD statement, which is similar to CLAIM checklist.Materials and MethodsOur systematic review was conducted in accordance with PRISMA guidelines. Ovid Embase, Ovid MEDLINE, Cochrane trials (CENTRAL) and Web of science core-collection were searched. Keywords included artificial intelligence, machine learning, deep learning, radiomics, magnetic resonance imaging, glioma, and glioblastoma. Assessment of TRIPOD adherence in 29 eligible studies was performed. Individual item performance was assessed by adherence index (ADI), the ratio of mean achieved score to maximum score per TRIPOD item.ResultsIn a preliminary analysis of 8 studies, the average TRIPOD adherence score was 0.48 (14.25/30 items fulfilled) with individual scores ranging from 0.27 (8/30) to 0.60 (18/30). Best overall item performance, with an ADI of 1, was seen in item 3 (Background/Objectives), 16 (Model performance) and 19 (Interpretation). Poorest performance was detected in item 1 (Title) and 2 (), followed by item 9 (Missing Data) with ADI of 0, 0 and 0.13, respectively.ConclusionPreliminary results underline the lack of reproducibility in ML studies on distinction between glioma and brain metastasis. An average TRIPOD adherence score of 0.48 indicates insufficient quality of reporting and outlines the need for increased utilization of quality scoring systems in study documentation. Systematic evaluation of quality score adherence will allow us to identify common flaws in this field for enabling translation of models into clinical workflow.

Review on Supervised and Unsupervised Learning Techniques for Electrical Power Systems: Algorithms and Applications

AbstractMachine learning (ML) has become a rising sophisticated technological application trend in the electrical industry in recent years. Such innovation provides optional methodologies for many existing applications, such as power and load profile forecasting, reliability evaluation, substation behavior detection and state observation of electrical equipment, and so on. This paper presents a review of various supervised and unsupervised ML techniques and applications for electrical power systems, including generation, transmission, distribution and micro‐grid. The algorithms and applications are mainly summarized from IEEE journals and the interest of this paper shows the roles and developments of most used algorithms and its corresponding extensions and performance in different applications. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Roles of PLCβ, PIP2 , and GIRK channels in arginine vasopressin-elicited excitation of CA1 pyramidal neurons.

Arginine vasopressin (AVP) is a hormone exerting vasoconstrictive and antidiuretic action in the periphery and serves as a neuromodulator in the brain. Althoughthe hippocampus receives vasopressinergic innervation and AVP has been shown to facilitate the excitability of CA1 pyramidal neurons, the involved ionic and signaling mechanisms have not been determined. Here we found that AVP excited CA1 pyramidal neurons by activation of V1a receptors. Functions of G proteins and phospholipase Cβ (PLCβ) were required for AVP-elicited excitation of CA1 pyramidal neurons, whereas intracellular Ca2+ release and protein kinase C were unnecessary. PLCβ-mediated depletion of phosphatidylinositol 4,5-bisphosphate (PIP2 ) was required for AVP-elicited excitation of CA1 pyramidal neurons. AVP augmented the input resistance and increased the time constants of CA1 pyramidal neurons. AVP induced an inward current in K+ -containing intracellular solution, whereas no inward currents were observed with Cs+ -containing intracellular solution. AVP-sensitive currents showed inward rectification with a reversal potential close to the K+ reversal potential, suggesting the involvement of inwardly rectifying K+ channels. AVP-induced currents were sensitive to the micromolar concentration of Ba2+ and tertiapin-Q, whereas application of ML 133, a selective Kir2 channel blocker had no effects, suggesting that AVP excited CA1 pyramidal neurons by depressing G protein-gated inwardly rectifying K+ channels. Activation of V1a receptors in the CA1 region facilitated glutamatergic transmission onto subicular pyramidal neurons, suggesting that AVP modulates network activity in the brain. Our results may provide one of the cellular and molecular mechanisms to explain the in vivo physiological functions of AVP.

A qualitative study of Machine Learning practices and engineering challenges in Earth Observation

Abstract Machine Learning (ML) is ubiquitously on the advance. Like many domains, Earth Observation (EO) also increasingly relies on ML applications, where ML methods are applied to process vast amounts of heterogeneous and continuous data streams to answer socially and environmentally relevant questions. However, developing such ML- based EO systems remains challenging: Development processes and employed workflows are often barely structured and poorly reported. The application of ML methods and techniques is considered to be opaque and the lack of transparency is contradictory to the responsible development of ML-based EO applications. To improve this situation a better understanding of the current practices and engineering-related challenges in developing ML-based EO applications is required. In this paper, we report observations from an exploratory study where five experts shared their view on ML engineering in semi-structured interviews. We analysed these interviews with coding techniques as often applied in the domain of empirical software engineering. The interviews provide informative insights into the practical development of ML applications and reveal several engineering challenges. In addition, interviewees participated in a novel workflow sketching task, which provided a tangible reflection of implicit processes. Overall, the results confirm a gap between theoretical conceptions and real practices in ML development even though workflows were sketched abstractly as textbook-like. The results pave the way for a large-scale investigation on requirements for ML engineering in EO.

Justicia: A Stochastic SAT Approach to Formally Verify Fairness

As a technology ML is oblivious to societal good or bad, and thus, the field of fair machine learning has stepped up to propose multiple mathematical definitions, algorithms, and systems to ensure different notions of fairness in ML applications. Given the multitude of propositions, it has become imperative to formally verify the fairness metrics satisfied by different algorithms on different datasets. In this paper, we propose a stochastic satisfiability (SSAT) framework, Justicia, that formally verifies different fairness measures of supervised learning algorithms with respect to the underlying data distribution. We instantiate Justicia on multiple classification and bias mitigation algorithms, and datasets to verify different fairness metrics, such as disparate impact, statistical parity, and equalized odds. Justicia is scalable, accurate, and operates on non-Boolean and compound sensitive attributes unlike existing distribution-based verifiers, such as FairSquare and VeriFair. Being distribution-based by design, Justicia is more robust than the verifiers, such as AIF360, that operate on specific test samples. We also theoretically bound the finite-sample error of the verified fairness measure.

Prediction of Cancer Disease using Machine learning Approach

Cancer has identified a diverse condition of several various subtypes. The timely screening and course of treatment of a cancer form is now a requirement in early cancer research because it supports the medical treatment of patients. Many research teams studied the application of ML and Deep Learning methods in the field of biomedicine and bioinformatics in the classification of people with cancer across high- or low-risk categories. These techniques have therefore been used as a model for the development and treatment of cancer. As, it is important that ML instruments are capable of detecting key features from complex datasets. Many of these methods are widely used for the development of predictive models for predicating a cure for cancer, some of the methods are artificial neural networks (ANNs), supportvector machine (SVMs) and decision trees (DTs). While we can understand cancer progression with the use of ML methods, an adequate validity level is needed to take these methods into consideration in clinical practice every day.In this study, the ML & DL approaches used in cancer progression modeling are reviewed. The predictions addressed are mostly linked to specific ML, input, and data samples supervision.

A review of predictive policing from the perspective of fairness

Machine Learning has become a popular tool in a variety of applications in criminal justice, including sentencing and policing. Media has brought attention to the possibility of predictive policing systems causing disparate impacts and exacerbating social injustices. However, there is little academic research on the importance of fairness in machine learning applications in policing. Although prior research has shown that machine learning models can handle some tasks efficiently, they are susceptible to replicating systemic bias of previous human decision-makers. While there is much research on fair machine learning in general, there is a need to investigate fair machine learning techniques as they pertain to the predictive policing. Therefore, we evaluate the existing publications in the field of fairness in machine learning and predictive policing to arrive at a set of standards for fair predictive policing. We also review the evaluations of ML applications in the area of criminal justice and potential techniques to improve these technologies going forward. We urge that the growing literature on fairness in ML be brought into conversation with the legal and social science concerns being raised about predictive policing. Lastly, in any area, including predictive policing, the pros and cons of the technology need to be evaluated holistically to determine whether and how the technology should be used in policing.

Model Inversion Attack: Analysis under Gray-box Scenario on Deep Learning based Face Recognition System

In a wide range of ML applications, the training data contains privacy-sensitive information that should be kept secure. Training the ML systems by privacy-sensitive data makes the ML model inherent to the data. As the structure of the model has been fine-tuned by training data, the model can be abused for accessing the data by the estimation in a reverse process called model inversion attack (MIA). Although, MIA has been applied to shallow neural network models of recognizers in literature and its threat in privacy violation has been approved, in the case of a deep learning (DL) model, its efficiency was under question. It was due to the complexity of a DL model structure, big number of DL model parameters, the huge size of training data, big number of registered users to a DL model and thereof big number of class labels. This research work first analyses the possibility of MIA on a deep learning model of a recognition system, namely a face recognizer. Second, despite the conventional MIA under the white box scenario of having partial access to the users' non-sensitive information in addition to the model structure, the MIA is implemented on a deep face recognition system by just having the model structure and parameters but not any user information. In this aspect, it is under a semi-white box scenario or in other words a gray-box scenario. The experimental results in targeting five registered users of a CNN-based face recognition system approve the possibility of regeneration of users' face images even for a deep model by MIA under a gray box scenario. Although, for some images the evaluation recognition score is low and the generated images are not easily recognizable, but for some other images the score is high and facial features of the targeted identities are observable. The objective and subjective evaluations demonstrate that privacy cyber-attack by MIA on a deep recognition system not only is feasible but also is a serious threat with increasing alert state in the future as there is considerable potential for integration more advanced ML techniques to MIA.

Machine learning: principles and applications for thoracic surgery.

Machine learning (ML) has experienced a revolutionary decade with advances across many disciplines. We seek to understand how recent advances in ML are going to specifically influence the practice of surgery in the future with a particular focus on thoracic surgery. Review of relevant literature in both technical and clinical domains. ML is a revolutionary technology that promises to change the way that surgery is practiced in the near future. Spurred by an advance in computing power and the volume of data produced in healthcare, ML has shown remarkable ability to master tasks that had once been reserved for physicians. Supervised learning, unsupervised learning and reinforcement learning are all important techniques that can be leveraged to improve care. Five key applications of ML to cardiac surgery include diagnostics, surgical skill assessment, postoperative prognostication, augmenting intraoperative performance and accelerating translational research. Some key limitations of ML include lack of interpretability, low quality and volumes of relevant clinical data, ethical limitations and difficulties with clinical implementation. In the future, the practice of cardiac surgery will be greatly augmented by ML technologies, ultimately leading to improved surgical performance and better patient outcomes.

Discovering and Dissecting Mechanically Excited Luminescence of Mn2+ Activators via Matrix Microstructure Evolution

AbstractMechanoluminescent (ML) materials featuring renewable mechanical‐to‐optical conversion have shown promising prospects in stress sensing, lighting, and display. However, the advancement in ML applications is being restrained by the obstacles in developing efficient ML materials and understanding the underlying ML mechanisms. Herein, a matrix evolution strategy to modulate the local microstructure and electronic environment around the luminescent activators is proposed, which not only supports the batch development of new ML materials but also provides a well‐connected platform for systematically revealing the mechanism of achieving efficient ML performance. The feasibility of the strategy is proved by constructing and evaluating a series of ML materials with matrix‐dependent luminescent properties in experimental‐theoretical collaboration. It is demonstrated that the construction of piezoluminescence is available in both non‐centrosymmetric and centrosymmetric matrices without being restricted by lattice symmetry. The inter‐electronic‐levels and shallow electron traps formed by activator doping enhance the electron recombination efficiency through tunneling and conduction band transfer pathways. The results are expected to accelerate the exploitation of ML material systems and to deepen the comprehensive apprehending of ML mechanisms, thereby guiding the rational design and widespread use of efficient ML materials.

AMELIORATION OF MACHINE LEARNING AND ARTIFICIAL INTELLIGENCE IN MEDICAL MANAGEMENT (A CASE STUDY ON THE PANDEMIC OF COVID 19 CASES IN ASIA)

Machine learning is seeing increasingly utilized in medical management for different reasons: tremendous sums of information are being captured and made accessible carefully; handling of vast sums of information has gotten to be cost-effective due to the expanded computing control presently accessible at reasonable costs; and different open source systems, toolkits, and libraries are accessible that can be utilized to construct and execute ML applications. Particularly in healthcare, ML has driven to energizing modern improvements that may rethink COVID-19 treatment and vaccination within a long time to come. With modern occurrences of the brand modern coronavirus clutter, COVID-19, creating day through the day, it is common to compare the unused affliction to other flare-ups in current history. Machine learning can offer assistance to anticipate three sorts of restorative dangers - disease, seriousness, and result. Whereas it is still it is early for COVID-19 with machine learning, but early applications seem promising. Machine learning is utilized when a computer has been instructed to recognize designs by giving it with information and a calculation to assist get it that information. This method of learning from the information is called training and the yield that we accomplish is through testing. Machine learning-based robotic surgery is changing the way surgery is performed nowadays. Machine learning in healthcare is getting to be more broadly utilized and is making a difference in patients and clinicians in numerous diverse ways. Cybersecurity and protection are major concerns and open challenges in healthcare which are yet to be addressed.

Modeling the Settling Velocity of a Sphere in Newtonian and Non-Newtonian Fluids with Machine-Learning Algorithms

The traditional procedure of predicting the settling velocity of a spherical particle is inconvenient as it involves iterations, complex correlations, and an unpredictable degree of uncertainty. The limitations can be addressed efficiently with artificial intelligence-based machine-learning algorithms (MLAs). The limited number of isolated studies conducted to date were constricted to specific fluid rheology, a particular MLA, and insufficient data. In the current study, the generalized application of ML was comprehensively investigated for Newtonian and three varieties of non-Newtonian fluids such as Power-law, Bingham, and Herschel Bulkley. A diverse set of nine MLAs were trained and tested using a large dataset of 967 samples. The ranges of generalized particle Reynolds number (ReG) and drag coefficient (CD) for the dataset were 10−3 < ReG (-) < 104 and 10−1 < CD (-) < 105, respectively. The performances of the models were statistically evaluated using an evaluation metric of the coefficient-of-determination (R2), root-mean-square-error (RMSE), mean-squared-error (MSE), and mean-absolute-error (MAE). The support vector regression with polynomial kernel demonstrated the optimum performance with R2 = 0.92, RMSE = 0.066, MSE = 0.0044, and MAE = 0.044. Its generalization capability was validated using the ten-fold-cross-validation technique, leave-one-feature-out experiment, and leave-one-data-set-out validation. The outcome of the current investigation was a generalized approach to modeling the settling velocity.

ROUND ABOUT MACHINE LEARNING

In this talk I will review the use of Machine Learning in studies of coastal morphodynamics. I will discuss a number of problems where ML tools have been used and why it makes sense to use ML methods. I will also outline recent advances in ML algorithms and applications, and discuss possible areas for future research.Recorded Presentation from the vICCE (YouTube Link): https://youtu.be/X5QnAdD1-T8