Machine Learning Experiments Research Articles

The Arc de Triomphe, Wrapped: Measuring Public Installation Art Engagement and Popularity through Social Media Data Analysis

Social media is the most popular canvas to engage with art. In this study, we provide a different angle, on how an artistic installation on a world-renowned monument—such as Paris’ Arc de Triomphe—can emotionally affect viewers and potentially increase the popularity of the artwork. We collected N = 7078 Instagram and N = 3776 Twitter posts of the Arc de Triomphe as wrapped (installation) and unwrapped using APIs. As engagement indicators for several supervised machine learning experiments, we chose the total number of likes, comments, shares, text sentiment, and so on. Our findings revealed that people were captivated by the poetic installation. Based on the results, we discovered that the sentiments of triumph and surprise prevailed in datasets of the Arc de Triomphe as unwrapped. The same sentiments of triumph and surprise were most prevalent in datasets as wrapped, as well, but with higher scores. Furthermore, we have provided evidence of public art experience and engagement in the social media era. This research, we believe, will be useful in future studies of social media through the lens of public art and popularity. We hope that our findings will stimulate future research in the fields of art curatorship, cultural heritage management, marketing and communication, aesthetics, and culture analytics.

Assessing the validity of test scores using response process data from an eye-tracking study: a new approach

Understanding the response process used by test takers when responding to multiple-choice questions (MCQs) is particularly important in evaluating the validity of score interpretations. Previous authors have recommended eye-tracking technology as a useful approach for collecting data on the processes test taker’s use to respond to test questions. This study proposes a new method for evaluating alternative score interpretations by using eye-tracking data and machine learning. We collect eye-tracking data from 26 students responding to clinical MCQs. Analysis is performed by providing 119 eye-tracking features as input for a machine-learning model aiming to classify correct and incorrect responses. The predictive power of various combinations of features within the model is evaluated to understand how different feature interactions contribute to the predictions. The emerging eye-movement patterns indicate that incorrect responses are associated with working from the options to the stem. By contrast, correct responses are associated with working from the stem to the options, spending more time on reading the problem carefully, and a more decisive selection of a response option. The results suggest that the behaviours associated with correct responses are aligned with the real-world model used for score interpretation, while those associated with incorrect responses are not. To the best of our knowledge, this is the first study to perform data-driven, machine-learning experiments with eye-tracking data for the purpose of evaluating score interpretation validity.

BotCap: Machine Learning Approach for Botnet Detection Based on Statistical Features

In this paper, we describe a detailed approach to develop a botnet detection system using machine learning (ML)techniques. Detecting botnet member hosts, or identifying botnet traffic has been the main subject of manyresearch efforts. This research aims to overcome two serious limitations of current botnet detection systems:First, the need for Deep Packet Inspection-DPI and the need to collect traffic from several infected hosts. Toachieve that, we have analyzed several botware samples of known botnets. Based on this analysis, we haveidentified a set of statistical features that may help to distinguish between benign and botnet malicious traffic.Then, we have carried several machine learning experiments in order to test the suitability of ML techniques andalso to pick a minimal subset of the identified features that provide best detection. We have implemented ourapproach in a tool called BotCap whose test results showed its proven ability to detect individually infected hostsin a local network.

Linguistic Parameters of Spontaneous Speech for Identifying Mild Cognitive Impairment and Alzheimer Disease

Abstract In this article, we seek to automatically identify Hungarian patients suffering from mild cognitive impairment (MCI) or mild Alzheimer disease (mAD) based on their speech transcripts, focusing only on linguistic features. In addition to the features examined in our earlier study, we introduce syntactic, semantic, and pragmatic features of spontaneous speech that might affect the detection of dementia. In order to ascertain the most useful features for distinguishing healthy controls, MCI patients, and mAD patients, we carry out a statistical analysis of the data and investigate the significance level of the extracted features among various speaker group pairs and for various speaking tasks. In the second part of the article, we use this rich feature set as a basis for an effective discrimination among the three speaker groups. In our machine learning experiments, we analyze the efficacy of each feature group separately. Our model that uses all the features achieves competitive scores, either with or without demographic information (3-class accuracy values: 68%–70%, 2-class accuracy values: 77.3%–80%). We also analyze how different data recording scenarios affect linguistic features and how they can be productively used when distinguishing MCI patients from healthy controls.

정적 영상 기반 행동 인식을 활용한 낙상 감지 기술 및 운전자 부주의 감지 기술 개발

Given image-level training data, a single test image can be used to identify a person’s actions or behaviors. This study highlights how the use of keypoint-based representation from single images can simplify the visual patterns of image events. Over the past few decades, deep learning techniques have successfully been applied in data-driven keypoint detection and skeletal analysis. While some engineered features from RGB or RGB-D datasets fail in image recognition applications due to ambiguous lightning conditions, previous knowledge from machine learning experimentation on large-scale data can be transferred into a new domain to improve performance. This idea of applying previously trained data can be applied to other applications as well. By adopting pre-trained convolutional neural network (CNN) models utilizing action recognition, new applications can be developed for events such as fall detection and driver drowsiness detection. Using state-of-the-art CNNs as a deep feature extractor to extract important key points of a human body or face, the geometric relationship of the predicted joints or facial features can be analyzed to aid in the design of hazardous event detection methods. The methods in this report were validated with publicly available datasets and successfully demonstrated in real-time. Two different data acquisition systems were used to train and validate these methods using real-world images and qualitatively verify them with a sequence of static images. Details of the algorithms and two practical applications are also outlined. The approach used in this study is scalable and can be extended to other hazardous event detection methods in the future.

Zoom gesture analysis for age-inappropriate internet content filtering

The Internet contains large amounts of adult content. With only a few taps, or mis-taps, an under-aged user can be exposed to age-inappropriate content. Currently, this can be avoided by creating age-restricted profiles or restricting users to child-friendly applications (apps). However, these existing measures are time-consuming, laborious, and require a higher level of technical literacy than many parents can afford. We believe a better solution is to use a browser or an app that automatically detects the user’s age then applies any appropriate content filters. For such a browser/app to be developed, we must learn that age estimation can indeed be performed with an acceptable rate of error. To that end, we created an Android app that collects biometric touchscreen data from elementary school, middle school, high school, and university students. Touch samples were collected from participants aged 5 to 61 on both smartphones and tablets. We focused exclusively on zoom-in and zoom-out touchscreen data samples. We made this decision because we found the zoom gesture to be rich with data and highly used among the most popular applications. Furthermore, we identify a niche within the current research landscape: no other machine learning experiments have leveraged the benefits of the zoom gesture for age estimation. We collected a total of 41,911 zoom data samples. From each zoom sample, 90 features were extracted. Those features were then used to train and test on six regressors and six classifiers to build a method that can accurately estimate the user’s age from their touchscreen behavior. The regressors performed with the best mean absolute errors (MAEs) of 2.27 and 2.54 years for smartphones and tablets, respectively. The classifiers performed with the best accuracies of 90% and 91% for smartphones and tablets, respectively. Given these results, it is our belief that not only is touch-based age estimation viable, but developing a child-safe browser or a parental control app with this underlying technology is a worthwhile endeavor.

Communication regarding the article "An efficient primary screening COVID-19 by serum Raman spectroscopy".

When performing computational modeling and machine learning experiments, it is imperative to follow a protocol that minimizes bias. In this communication, we share our concerns regarding the article “An efficient primary screening COVID‐19 by serum Raman spectroscopy” published in this journal. We consider that the authors may have inadvertently biased their results by not guaranteeing complete independence of test samples from the training data. We corroborate our point by reproducing the experiment with the available data, showing that if full independence of the test set was ensured, the reported results should be lower. We ask the authors to provide more information regarding their article, as well as making available all code used to generate their results. Our experiments are available at https://doi.org/10.6084/m9.figshare.14124356.

Data-Driven Design and Autonomous Experimentation in Soft and Biological Materials Engineering.

This article reviews recent developments in the applications of machine learning, data-driven modeling, transfer learning, and autonomous experimentation for the discovery, design, and optimization of soft and biological materials. The design and engineering of molecules and molecular systems have long been a preoccupation of chemical and biomolecular engineers using a variety of computational and experimental techniques. Increasingly, researchers have looked to emerging and established tools in artificial intelligence and machine learning to integrate with established approaches in chemical science to realize powerful, efficient, and in some cases autonomous platforms for molecular discovery, materials engineering, and process optimization. This review summarizes the basic principles underpinning these techniques and highlights recent successful example applications in autonomous materials discovery, transfer learning, and multi-fidelity active learning.

Discovery of Kinase and Carbonic Anhydrase Dual Inhibitors by Machine Learning Classification and Experiments.

A multi-target small molecule modulator is advantageous for treating complicated diseases such as cancers. However, the strategy and application for discovering a multi-target modulator have been less reported. This study presents the dual inhibitors for kinase and carbonic anhydrase (CA) predicted by machine learning (ML) classifiers, and validated by biochemical and biophysical experiments. ML trained by CA I and CA II inhibitor molecular fingerprints predicted candidates from the protein-specific bioactive molecules approved or under clinical trials. For experimental tests, three sulfonamide-containing kinase inhibitors, 5932, 5946, and 6046, were chosen. The enzyme assays with CA I, CA II, CA IX, and CA XII have allowed the quantitative comparison in the molecules’ inhibitory activities. While 6046 inhibited weakly, 5932 and 5946 exhibited potent inhibitions with 100 nM to 1 μM inhibitory constants. The ML screening was extended for finding CAs inhibitors of all known kinase inhibitors. It found XMU-MP-1 as another potent CA inhibitor with an approximate 30 nM inhibitory constant for CA I, CA II, and CA IX. Differential scanning fluorimetry confirmed the direct interaction between CAs and small molecules. Cheminformatics studies, including docking simulation, suggest that each molecule possesses two separate functional moieties: one for interaction with kinases and the other with CAs.

Explainable artificial intelligence to predict the risk of side-specific extraprostatic extension in pre-prostatectomy patients.

We aimed to develop an explainable machine learning (ML) model to predict side-specific extraprostatic extension (ssEPE) to identify patients who can safely undergo nerve-sparing radical prostatectomy using preoperative clinicopathological variables. A retrospective sample of clinicopathological data from 900 prostatic lobes at our institution was used as the training cohort. Primary outcome was the presence of ssEPE. The baseline model for comparison had the highest performance out of current biopsy-derived predictive models for ssEPE. A separate logistic regression (LR) model was built using the same variables as the ML model. All models were externally validated using a testing cohort of 122 lobes from another institution. Models were assessed by area under receiver-operating-characteristic curve (AUROC), precision-recall curve (AUPRC), calibration, and decision curve analysis. Model predictions were explained using SHapley Additive exPlanations. This tool was deployed as a publicly available web application. Incidence of ssEPE in the training and testing cohorts were 30.7 and 41.8%, respectively. The ML model achieved AUROC 0.81 (LR 0.78, baseline 0.74) and AUPRC 0.69 (LR 0.64, baseline 0.59) on the training cohort. On the testing cohort, the ML model achieved AUROC 0.81 (LR 0.76, baseline 0.75) and AUPRC 0.78 (LR 0.75, baseline 0.70). The ML model was explainable, well-calibrated, and achieved the highest net benefit for clinically relevant cutoffs of 10-30%. We developed a user-friendly application that enables physicians without prior ML experience to assess ssEPE risk and understand factors driving these predictions to aid surgical planning and patient counselling (https://share.streamlit.io/jcckwong/ssepe/main/ssEPE_V2.py).

On Multi-Agent Cognitive Cooperation: Can virtual agents behave like humans?

Individuals tend to cooperate or collaborate to reach a common goal when the going gets tough creating a common frame of reference that is a common mental representation of the situation. Information exchange among people is fundamental for building a shared strategy through the grounding process that exploits different communication channels like vision, haptic, or voice. Indeed, human perception is typically multi-modal.This work proposes a two-fold study investigating the cognitive collaboration process both among humans and virtual agents of a multi-agent reinforcement learning (MARL) system. The experiment with humans consists of an interactive virtual shared environment that uses multi-modal channels (visual and haptics) as interaction cues. Haptic feedback is fundamental for a good sense of presence and for improving the performance in completing a task. In this manuscript, an experiment, consisting of escaping a virtual maze trying to get the best score possible, is introduced. The experiment is meant to be performed in pairs, and the perceptual information is split among the participants. A custom haptic interface has been used for the interaction with the virtual environment. The machine learning case, instead, proposes two virtual agents implemented using a tabular Q-learning paradigm to control a single avatar in a 2D labyrinth, introducing a new form of MARL setting.As it is known, it is not easy to get familiar with haptics for people that have never used it, and that if not properly transmitted, the cognitive workflow does not produce any improvements. However, the main findings of the proposed work are that haptic-driven multi-modal feedback information is a valuable means of collaboration since it allows to establish a common frame of reference between the two participants. The machine learning experiments show that even independent agents, implemented with properly designed rewards, can learn the intentions of the other participant in the same environment and collaborate to accomplish a common task.

Exceptionally durable and hard metal alloys discovered with combination technique

By exploiting the synergy between machine learning and high throughput experimentation, scientists discovered exceptional metallic glass alloys.

Cfcv - Statistical Data-Driven Approach Towards Participation, Inclusion and Reciprocity in Machine Learning Experiment

Cfcv - Statistical Data-Driven Approach Towards Participation, Inclusion and Reciprocity in Machine Learning Experiment

New venues in electron density analysis

We provide a comprehensive overview of the chemical information from electron density: not only how to extract information, but also how to obtain and how to assess the quality of the electron density itself. After introducing several indexes derived from electron density, which allow bonding to be revealed, we focus on the various potential sources of electron density, and also explain the error trends they show so that a judicious choice of methods and limitations are clearly laid on the table. Computational, experimental-computational combinations, and machine learning efforts are covered in this work.

Quantum Kernels for Real-World Predictions Based on Electronic Health Records

Research on near-term quantum machine learning has explored how classical machine learning algorithms endowed with access to quantum kernels (similarity measures) can outperform their purely classical counterparts. Although theoretical work has shown provable advantage on synthetic data sets, no work done to date has studied empirically whether quantum advantage is attainable and with what data. In this paper, we report the first systematic investigation of empirical quantum advantage (EQA) in healthcare and life sciences and propose an end-to-end framework to study EQA. We selected electronic health records (EHRs) data subsets and created a configuration space of 5-20 features and 200-300 training samples. For each configuration coordinate, we trained classical support vector machine (SVM) models based on radial basis function (RBF) kernels and quantum models with custom kernels using an IBM quantum computer, making this one of the largest quantum machine learning experiments to date. We empirically identified regimes where quantum kernels could provide advantage and introduced a terrain ruggedness index, a metric to help quantitatively estimate how the accuracy of a given model will perform. The generalizable framework introduced here represents a key step towards a priori identification of data sets where quantum advantage could exist.

A machine learning algorithm to analyse the effects of vaccination on COVID-19 mortality.

The coronavirus disease 2019 (COVID-19), with new variants, continues to be a constant pandemic threat that is generating socio-economic and health issues in manifold countries. The principal goal of this study is to develop a machine learning experiment to assess the effects of vaccination on the fatality rate of the COVID-19 pandemic. Data from 192 countries are analysed to explain the phenomena under study. This new algorithm selected two targets: the number of deaths and the fatality rate. Results suggest that, based on the respective vaccination plan, the turnout in the participation in the vaccination campaign, and the doses administered, countries under study suddenly have a reduction in the fatality rate of COVID-19 precisely at the point where the cut effect is generated in the neural network. This result is significant for the international scientific community. It would demonstrate the effective impact of the vaccination campaign on the fatality rate of COVID-19, whatever the country considered. In fact, once the vaccination has started (for vaccines that require a booster, we refer to at least the first dose), the antibody response of people seems to prevent the probability of death related to COVID-19. In short, at a certain point, the fatality rate collapses with increasing doses administered. All these results here can help decisions of policymakers to prepare optimal strategies, based on effective vaccination plans, to lessen the negative effects of the COVID-19 pandemic crisis in socioeconomic and health systems.

A Machine Learning Approach for the Classification of Falls and Activities of Daily Living in Agricultural Workers

Population aging is a global trend, and the highest proportion of elderly people in the workforce per unit of population is found in agricultural areas. However, few systematic studies have been conducted on farmer falls in the field of agricultural machinery. This study focuses on the application of classification methods for monitoring devices to detect fall/nonfall movements of farmworkers, where agricultural biomechanical factors are considered in detecting activities of daily living. In this study, we recorded and analyzed original acquisition datasets of signals obtained from two accelerometers and one gyroscope for 40 healthy individuals who performed various falls and activities of daily living (ADLs). Spatial characteristics were used to train the machine-learning classifiers to distinguish between fall and non-fall events. Supervised machine learning experiments evaluated the effectiveness of the proposed approach: the k-nearest neighbors (kNN) and support vector machine (SVM) algorithms achieved roc auc-scores of 0.999 in differentiating between falls and ADLs (binary-class classification). Moreover, an artificial neural network (ANN) classifier showed the highest performance in terms of classification roc auc-scores of 1.0. The evaluation metric demonstrated the highest performance in the analysis and evaluation of the signal obtained from the S2 acceleration sensor with a measurement range of ±16 g. The proposed SVM classifier evaluations showed a 0.988 roc auc-score for sensor tests in multi-class classification, along with the highest performance in terms of the F1-score and Matthews Correlation Coefficient (MCC) over 84% in the multi-class classification model for distinguishing each of ADLs and Fall using ±16 g acceleration sensor.

COVID and nutrition: A machine learning perspective

A self-report questionnaire survey was conducted online to collect big data from over 16000 Iranian families (who were the residents of 1000 urban and rural areas of Iran). The resulting data storage contained over 1 M records of data and over 1G records of automatically inferred information. Based on this data storage, a series of machine learning experiments was conducted to investigate the relationship between nutrition and the risk of contracting COVID-19. With highly accurate scores, the findings strongly suggest that foods and water sources containing certain natural bioactive and phytochemical agents may help to reduce the risk of apparent COVID-19 infection.

Evolution of single-lead ECG for STEMI detection using a deep learning approach

BackgroundWhile ST-Elevation Myocardial Infarction (STEMI) door-to-balloon times are often below 90 min, symptom to door times remain long at 2.5-h, due at least in part to a delay in diagnosis. ObjectivesTo develop and validate a machine learning-guided algorithm which uses a single‑lead electrocardiogram (ECG) for STEMI detection to speed diagnosis. MethodsData was extracted from the Latin America Telemedicine Infarct Network (LATIN), a population-based Acute Myocardial Infarction (AMI) program that provides care to patients in Brazil, Colombia, Mexico, and Argentina through telemedicine. Sample: the first dataset was comprised of 8511 ECGs that were used for various machine learning experiments to test our Deep Learning approach for STEMI diagnosis. The second dataset of 2542 confirmed STEMI diagnosis EKG records, including specific ischemic heart wall information (anterior, inferior, and lateral), was derived from the previous dataset to test the STEMI localization model. Preprocessing: Detection of QRS complexes by wavelet system, segmentation of each EKG record into individual heartbeats with fixed window of 0.4 s to the left and 0.9 s to the right of main. Training & Testing: 90% and 10% of the total dataset, respectively, were used for both models. Classification: two 1-D convolutional neural networks were implemented, two classes were considered for first models (STEMI/Not-STEMI) and three classes for the second model (Anterior/Inferior/Lateral) each corresponding to the heart wall affected. These individual probabilities were aggregated to generate the final label for each model. ResultsThe single‑lead ECG strategy was able to provide an accuracy of 90.5% for STEMI detection with Lead V2, which also yielded the best results overall among individual leads. STEMI Localization model provided promising results for anterior and inferior wall STEMIs but remained suboptimal for Lateral STEMI. ConclusionsAn Artificial Intelligence-enhanced single‑lead ECG is a promising screening tool. This technology provides an autonomous and accurate STEMI diagnostic alternative that can be incorporated into wearable devices, potentially providing patients reliable means to seek treatment early and offers the potential to thereby improve STEMI outcomes in the long run.

Semantic Description of Explainable Machine Learning Workflows for Improving Trust

Explainable Machine Learning comprises methods and techniques that enable users to better understand the machine learning functioning and results. This work proposes an ontology that represents explainable machine learning experiments, allowing data scientists and developers to have a holistic view, a better understanding of the explainable machine learning process, and to build trust. We developed the ontology by reusing an existing domain-specific ontology (ML-SCHEMA) and grounding it in the Unified Foundational Ontology (UFO), aiming at achieving interoperability. The proposed ontology is structured in three modules: (1) the general module, (2) the specific module, and (3) the explanation module. The ontology was evaluated using a case study in the scenario of the COVID-19 pandemic using healthcare data from patients, which are sensitive data. In the case study, we trained a Support Vector Machine to predict mortality of patients infected with COVID-19 and applied existing explanation methods to generate explanations from the trained model. Based on the case study, we populated the ontology and queried it to ensure that it fulfills its intended purpose and to demonstrate its suitability.