Binary Machine Research Articles

UniBin: Assembly semantic-enhanced binary vulnerability detection without disassembly

The widespread reuse of open-source code amplifies the impact of vulnerabilities. Current vulnerability detection methods predominantly rely on binary code similarity comparisons, which involve disassembling to obtain assembly code or control flow graphs. These methods depend on specific disassembly tools and complex preprocessing, limiting their applicability and detection speed. This paper proposes UniBin, a vulnerability detection method based on the multi-layer Transformer encoder. By employing bidirectional LM, unidirectional LM, and sequence-to-sequence LM tasks on both binary and assembly code during the pre-training phase, UniBin learns richer semantic information from binary machine code, enabling efficient similarity comparison without disassembly and mitigating the limitations of disassembly. We cross-compile 55 widely used open-source C projects as datasets. After 52 hours of pre-training and 8 hours of fine-tuning, UniBin reaches an average accuracy of 98.3% in similarity detection across compilation conditions, outperforming the state-of-the-art method. For search tasks across optimization options with a pool size of 1000, the Recall@1 metric improves by 28.2% (from 67.9% to 87.1%). UniBin eliminates dependency on specific disassembly tools and improves end-to-end binary analysis speed by over 36%. In real-world vulnerability detection tasks, UniBin detects all vulnerability functions with the lowest false positive rate of 0.16%.

Bayesian real-time classification of multi-messenger electromagnetic and gravitational-wave observations

Because of the electromagnetic (EM) radiation produced during the merger, compact binary coalescences with neutron stars may result in multi-messenger observations. In order to follow up on the gravitational-wave (GW) signal with EM telescopes, it is critical to promptly identify the properties of these sources. This identification must rely on the properties of the progenitor source, such as the component masses and spins, as determined by low-latency detection pipelines in real time. The output of these pipelines, however, might be biased, which could decrease the accuracy of parameter recovery. Machine learning algorithms are used to correct this bias. In this work, we revisit this problem and discuss two new implementations of supervised machine learning algorithms, K-nearest neighbors and random forest, which are able to predict the presence of a neutron star and post-merger matter remnant in low-latency compact binary coalescence searches across different search pipelines and data sets. Additionally, we present a novel approach for calculating the Bayesian probabilities for these two metrics. Instead of metric scores derived from binary machine learning classifiers, our scheme is designed to provide the astronomy community well-defined probabilities. This would deliver a more direct and easily interpretable product to assist EM telescopes in deciding whether to follow up on GW events in real time.

Can You Give Me a Lift?

Static Analysis (SA) is the practice of examining computer program source code for errors or vulnerabilities outside the compiler’s capabilities. To carry out Static Analysis on computer programs, various tools exist that parse and examine each line for known issues. These tools do not compile the program, nor do they run the program. Instead, they analyse the source code directly and infer properties about the program without executing it - thus “static” analysis. Static analysis is not new. Early UNIX contained a program called “lint” for static analysis; a tool that has now existed since the late 1970s (Johnson 1978). Increasingly, however, modern static analysis practices indicate a more focused intent; find cybersecurity weaknesses. At the 22nd European Conference on Cyber Warfare and Security, the authors presented the background of Intermediate Representations (IRs) and described how this “middleware” representation can be utilized for static analysis of source code for cybersecurity weaknesses. By examining an IR, potential flaws in the source code can be located. When utilizing the IR as opposed to the original high-level language, the static analysis process becomes independent of the original source language; if several languages such as C, Rust, and others all compile into the same IR, static analysis of the IR allows the analysis process to no longer be tied to the high-level language grammar or syntax. The previous paper implemented a literature survey of available IR analysis tools to discover prior work; the authors have subsequently advanced the research and are actively using an IR framework, LLVM, for vulnerability analysis. In this research, source code in a high-level language is first compiled to LLVM and the resulting IR is used for analysis. This approach uses a “code-to-IR” SA analysis preparation paradigm. At the same time, there is the potential for binary “lifters” to be used. These tools “lift” an executable program – binary machine instructions – back to LLVM. In this way, the paradigm can also be reversed such that static analysis of LLVM can be performed on source code compiled into LLVM, or on executable programs in the field that are “lifted”. This begs: how effective are these “lifters”? In this work, the authors present experiences in installation and operation of several binary “lifters” available as open-source projects. Some are supported better than others, some operate better than others, and some don’t operate at all. Those that do lead to the follow-up: Is the “lifted” code suitable for static analysis, or is it too obfuscated relative to the original program? This paper describes just that – our efforts and results in locating a binary “lifter” suitable for bringing executable program test cases back to LLVM for analysis by the cybersecurity vulnerability tool concurrently under development.

Speech features-based Parkinson’s disease classification using combined SMOTE-ENN and binary machine learning

Speech features-based Parkinson’s disease classification using combined SMOTE-ENN and binary machine learning

Feature Engineering and Decision Trees for Predicting High Crash-Risk Locations Using Roadway Indicators

Road crashes are a prevalent public health issue across the globe. The objective of this research was to develop a methodology for accurately classifying high-risk crash locations. The hypothesis of this study was that readily obtained roadway indicators can be used along with machine learning techniques to categorize locations as high crash-risk. A database containing 5,383 locations was created during 2012 to 2015 as part of the Hellenic National Road Safety Project and used to develop three binary machine learning models to classify high crash-risk locations based on roadway indicators. The three models were random forest, gradient boosting, and extra trees. This research used features engineering to reduce the number of indicators in the model, and the synthetic minority oversampling technique to address imbalances in the dataset between the minority (high crash-risk locations identified using crash reports) and majority classes (medium to low crash-risk locations identified based on local police testimonies, site inspections, and geometry analysis). Although all three models performed similarly, the extra trees model outperformed the other two on a range of performance metrics, including the area under the precision–recall curve and the F1-score. The findings revealed that design speeds, pavement markings, signage presence, and pavement condition were the most influential factors affecting roadway safety. The contribution of this research is in the development of a transferable methodology for classifying high crash-risk locations in addition to revealing key indicators for crash-risk potential, which in turn can inform cost-effective data collection and maintenance activities.

Loading pattern optimization using simulated annealing and binary machine learning pre-screening

We introduce a creative approach combining machine learning with optimization techniques to enhance the optimization of the loading pattern (LP). Finding the optimal LP is a critical decision that impacts both the reload safety and the economic feasibility of the nuclear fuel cycle. While simulated annealing (SA) is a widely accepted technique to solve the LP optimization problem, it suffers from the drawback of high computational cost since LP optimization requires three-dimensional depletion calculations. In this note, we introduce a technique to tackle this issue by leveraging neural networks to filter out inappropriate patterns, thereby reducing the number of SA evaluations. We demonstrate the efficacy of our novel approach by constructing a machine learning-based optimization model for the LP data of the Korea Standard Nuclear Power Plant (OPR-1000).

Radiomics of pericardial fat: a new frontier in heart failure discrimination and prediction

ObjectivesTo use pericardial adipose tissue (PAT) radiomics phenotyping to differentiate existing and predict future heart failure (HF) cases in the UK Biobank.MethodsPAT segmentations were derived from cardiovascular magnetic resonance (CMR) studies using an automated quality-controlled model to define the region-of-interest for radiomics analysis. Prevalent (present at time of imaging) and incident (first occurrence after imaging) HF were ascertained using health record linkage. We created balanced cohorts of non-HF individuals for comparison. PyRadiomics was utilised to extract 104 radiomics features, of which 28 were chosen after excluding highly correlated ones (0.8). These features, plus sex and age, served as predictors in binary classification models trained separately to detect (1) prevalent and (2) incident HF. We tested seven modeling methods using tenfold nested cross-validation and examined feature importance with explainability methods.ResultsWe studied 1204 participants in total, 297 participants with prevalent (60 ± 7 years, 21% female) and 305 with incident (61 ± 6 years, 32% female) HF, and an equal number of non-HF comparators. We achieved good discriminative performance for both prevalent (voting classifier; AUC: 0.76; F1 score: 0.70) and incident (light gradient boosting machine: AUC: 0.74; F1 score: 0.68) HF. Our radiomics models showed marginally better performance compared to PAT area alone.Increased PAT size (maximum 2D diameter in a given column or slice) and texture heterogeneity (sum entropy) were important features for prevalent and incident HF classification models.ConclusionsThe amount and character of PAT discriminate individuals with prevalent HF and predict incidence of future HF.Clinical relevance statementThis study presents an innovative application of pericardial adipose tissue (PAT) radiomics phenotyping as a predictive tool for heart failure (HF), a major public health concern. By leveraging advanced machine learning methods, the research uncovers that the quantity and characteristics of PAT can be used to identify existing cases of HF and predict future occurrences. The enhanced performance of these radiomics models over PAT area alone supports the potential for better personalised care through earlier detection and prevention of HF.Key Points•PAT radiomics applied to CMR was used for the first time to derive binary machine learning classifiers to develop models for discrimination of prevalence and prediction of incident heart failure.•Models using PAT area provided acceptable discrimination between cases of prevalent or incident heart failure and comparator groups.•An increased PAT volume (increased diameter using shape features) and greater texture heterogeneity captured by radiomics texture features (increased sum entropy) can be used as an additional classifier marker for heart failure.

Machine learning software to learn negligible elements of the Hamiltonian matrix

As a follow-up to our recent Communication in the Journal of Chemical Physics [J. Chem. Phys. 159 071101 (2023)], we report and make available the Jupyter Notebook software here. This software performs binary machine learning classification (MLC) with the goal of learning negligible Hamiltonian matrix elements for vibrational dynamics. We illustrate its usefulness for a Hamiltonian matrix for H2O by using three MLC algorithms: Random Forest, Support Vector Machine, and Multi-layer Perceptron.

RNAinsecta: A tool for prediction of precursor microRNA in insects and search for their target in the model organism Drosophila melanogaster.

Pre-MicroRNAs are the hairpin loops from which microRNAs are produced that have been found to negatively regulate gene expression in several organisms. In insects, microRNAs participate in several biological processes including metamorphosis, reproduction, immune response, etc. Numerous tools have been designed in recent years to predict novel pre-microRNA using binary machine learning classifiers where prediction models are trained with true and pseudo pre-microRNA hairpin loops. Currently, there are no existing tool that is exclusively designed for insect pre-microRNA detection. Application of machine learning algorithms to develop an open source tool for prediction of novel precursor microRNA in insects and search for their miRNA targets in the model insect organism, Drosophila melanogaster. Machine learning algorithms such as Random Forest, Support Vector Machine, Logistic Regression and K-Nearest Neighbours were used to train insect true and false pre-microRNA features with 10-fold Cross Validation on SMOTE and Near-Miss datasets. miRNA targets IDs were collected from miRTarbase and their corresponding transcripts were collected from FlyBase. We used miRanda algorithm for the target searching. In our experiment, SMOTE performed significantly better than Near-Miss for which it was used for modelling. We kept the best performing parameters after obtaining initial mean accuracy scores >90% of Cross Validation. The trained models on Support Vector Machine achieved accuracy of 92.19% while the Random Forest attained an accuracy of 80.28% on our validation dataset. These models are hosted online as web application called RNAinsecta. Further, searching target for the predicted pre-microRNA in Drosophila melanogaster has been provided in RNAinsecta.

A machine learning-based assistant tool for early frailty screening of patients receiving maintenance hemodialysis.

To develop an assistant tool based on machine learning for early frailty screening in patients receiving maintenance hemodialysis. This is a single-center retrospective study. 141 participants' basic information, scale results and laboratory findings were collected and the FRAIL scale was used to assess frailty. Then participants were divided into the frailty group (n = 84) and control group (n = 57). After feature selection, data split and oversampling, ten commonly used binary machine learning methods were performed and a voting classifier was developed. The grade results of Clinical Frailty Scale, age, serum magnesium, lactate dehydrogenase, comorbidity and fast blood glucose were considered to be the best feature set for early frailty screening. After abandoning models with overfitting or poor performance, the voting classifier based on Support Vector Machine, Adaptive Boosting and Naive Bayes achieved a good screening performance (sensitivity: 68.24% ± 8.40%, specificity:72.50% ± 11.81%, F1 score: 72.55% ± 4.65%, AUC:78.38% ± 6.94%). A simple and efficient early frailty screening assistant tool for patients receiving maintenance hemodialysis based on machine learning was developed. It can provide assistance on frailty, especially pre-frailty screening and decision-making tasks.

DNA methylation signature classification of rare disorders using publicly available methylation data.

Disease-specific DNA methylation patterns (DNAm signatures) have been established for an increasing number of genetic disorders and represent a valuable tool for classification of genetic variants of uncertain significance (VUS). Sample size and batch effects are critical issues for establishing DNAm signatures, but their impact on the sensitivity and specificity of an already established DNAm signature has not previously been tested. Here, we assessed whether publicly available DNAm data can be employed to generate a binary machine learning classifier for VUS classification, and used variants in KMT2D, the gene associated with Kabuki syndrome, together with an existing DNAm signature as proof-of-concept. Using publicly available methylation data for training, a classifier for KMT2D variants was generated, and individuals with molecularly confirmed Kabuki syndrome and unaffected individuals could be correctly classified. The present study documents the clinical utility of a robust DNAm signature even for few affected individuals, and most importantly, underlines the importance of data sharing for improved diagnosis of rare genetic disorders.

External validation of binary machine learning models for pain intensity perception classification from EEG in healthy individuals

Discrimination of pain intensity using machine learning (ML) and electroencephalography (EEG) has significant potential for clinical applications, especially in scenarios where self-report is unsuitable. However, existing research is limited due to a lack of external validation (assessing performance using novel data). We aimed for the first external validation study for pain intensity classification with EEG. Pneumatic pressure stimuli were delivered to the fingernail bed at high and low pain intensities during two independent EEG experiments with healthy participants. Study one (n = 25) was utilised for training and cross-validation. Study two (n = 15) was used for external validation one (identical stimulation parameters to study one) and external validation two (new stimulation parameters). Time–frequency features of peri-stimulus EEG were computed on a single-trial basis for all electrodes. ML training and analysis were performed on a subset of features, identified through feature selection, which were distributed across scalp electrodes and included frontal, central, and parietal regions. Results demonstrated that ML models outperformed chance. The Random Forest (RF) achieved the greatest accuracies of 73.18, 68.32 and 60.42% for cross-validation, external validation one and two, respectively. Importantly, this research is the first to externally validate ML and EEG for the classification of intensity during experimental pain, demonstrating promising performance which generalises to novel samples and paradigms. These findings offer the most rigorous estimates of ML’s clinical potential for pain classification.

OA-Pain-Sense: Machine Learning Prediction of Hip and Knee Osteoarthritis Pain from IMU Data

Joint pain is a prominent symptom of Hip and Knee Osteoarthritis (OA), impairing patients’ movements and affecting the joint mechanics of walking. Self-report questionnaires are currently the gold standard for Hip OA and Knee OA pain assessment, presenting several problems, including the fact that older individuals often fail to provide accurate self-pain reports. Passive methods to assess pain are desirable. This study aims to explore the feasibility of OA-Pain-Sense, a passive, automatic Machine Learning-based approach that predicts patients’ self-reported pain levels using SpatioTemporal Gait features extracted from the accelerometer signal gathered from an anterior-posterior wearable sensor. To mitigate inter-subject variability, we investigated two types of data rescaling: subject-level and dataset-level. We explored six different binary machine learning classification models for discriminating pain in patients with Hip OA or Knee OA from healthy controls. In rigorous evaluation, OA-Pain-Sense achieved an average accuracy of 86.79% using the Decision Tree and 83.57% using Support Vector Machine classifiers for distinguishing Hip OA and Knee OA patients from healthy subjects, respectively. Our results demonstrate that OA-Pain-Sense is feasible, paving the way for the development of a pain assessment algorithm that can support clinical decision-making and be used on any wearable device, such as smartphones.

Ensemble Feature Selection in Binary Machine Learning Classification: A Novel Application of the Evaluation Based on Distance from Average Solution (EDAS) Method

Combining filters in an ensemble to improve feature selection performance is a growing field in the literature. Current techniques, however, are focused on approaches that suffer from drawbacks such as sensitivity to skewed distribution, among others. To address this gap, this paper investigates the applicability of multiple criteria decision-making in ensemble feature selection. This paper adopts the Evaluation based on Distance from Average Solution (EDAS) method due to its many familiar elements to the feature selection community. An experiment was performed on six datasets and a control group. The paper uses the six datasets as levels of the blocking factor. A negative control group (i.e., no feature selection) was adopted to compare with the proposed algorithm. Results show that the proposed ensemble FS algorithm was able to reduce the dataset without compromising the performance of the classifier. The findings in this study would contribute to the literature in several ways. First, the paper is one of the few works to demonstrate how MCDM can be used in feature selection with promising results. Second, this paper is one of the few works to demonstrate the significance of including datasets as levels of a blocking factor when performing significance testing. Finally, this paper is the first to demonstrate the applicability of EDAS as an ensemble FS algorithm. As such, the findings in this paper could spark the cross-fertilization of feature selection and MCDM.

Artificial intelligence in differentiating tropical infections: A step ahead.

Background and objectiveDifferentiating tropical infections are difficult due to its homogenous nature of clinical and laboratorial presentations among them. Sophisticated differential tests and prediction tools are better ways to tackle this issue. Here, we aimed to develop a clinician assisted decision making tool to differentiate the common tropical infections.MethodologyA cross sectional study through 9 item self-administered questionnaire were performed to understand the need of developing a decision making tool and its parameters. The most significant differential parameters among the identified infections were measured through a retrospective study and decision tree was developed. Based on the parameters identified, a multinomial logistic regression model and a machine learning model were developed which could better differentiate the infection.ResultsA total of 40 physicians involved in the management of tropical infections were included for need analysis. Dengue, malaria, leptospirosis and scrub typhus were the common tropical infections in our settings. Sodium, total bilirubin, albumin, lymphocytes and platelets were the laboratory parameters; and abdominal pain, arthralgia, myalgia and urine output were the clinical presentation identified as better predictors. In multinomial logistic regression analysis with dengue as a reference revealed a predictability of 60.7%, 62.5% and 66% for dengue, malaria and leptospirosis, respectively, whereas, scrub typhus showed only 38% of predictability. The multi classification machine learning model observed to have an overall predictability of 55–60%, whereas a binary classification machine learning algorithms showed an average of 79–84% for one vs other and 69–88% for one vs one disease category.ConclusionThis is a first of its kind study where both statistical and machine learning approaches were explored simultaneously for differentiating tropical infections. Machine learning techniques in healthcare sectors will aid in early detection and better patient care.

ROC Curves, Loss Functions, and Distorted Probabilities in Binary Classification

The main purpose of this work is to study how loss functions in machine learning influence the “binary machines”, i.e., probabilistic AI models for predicting binary classification problems. In particular, we show the following results: (i) Different measures of accuracy such as area under the curve (AUC) of the ROC curve, the maximal balanced accuracy, and the maximally weighted accuracy are topologically equivalent, with natural inequalities relating them; (ii) the so-called real probability machines with respect to given information spaces are the optimal machines, i.e., they have the highest precision among all possible machines, and moreover, their ROC curves are automatically convex; (iii) the cross-entropy and the square loss are the most natural loss functions in the sense that the real probability machine is their minimizer; (iv) an arbitrary strictly convex loss function will also have as its minimizer an optimal machine, which is related to the real probability machine by just a reparametrization of sigmoid values; however, if the loss function is not convex, then its minimizer is not an optimal machine, and strange phenomena may happen.

Biochemical algorithm to identify individuals with ALPL variants among subjects with persistent hypophosphatasaemia

BackgroundHypophosphatasia (HPP) is a rare and underdiagnosed condition characterized by deficient bone and teeth mineralization. The aim of this study was first, to evaluate the diagnostic utility of employing alkaline phosphatase (ALP) threshold levels to identify adults with variants in ALPL among individuals with persistently low ALP levels and second, to determine the value of also including its substrates (serum pyridoxal-5′-phosphate—PLP—and urinary phosphoetanolamine-PEA) for this purpose in order to create a biochemical algorithm that could facilitate the diagnostic work-up of HPP.ResultsThe study population comprised 77 subjects with persistent hypophosphatasaemia. They were divided into two groups according to the presence (+GT) or absence (−GT) of pathogenic ALPL variants: 40 +GT and 37 −GT. Diagnostic utility measures were calculated for different ALP thresholds and Receiver Operating Characteristic (ROC) curves were employed to determine PLP and PEA optimal cut-off levels to predict the presence of variants. The optimal threshold for ALP was 25 IU/L; for PLP, 180 nmol/L and for PEA, 30 µmol/g creatinine. Biochemical predictive models were assessed using binary logistic regression analysis and bootstrapping machine learning technique and results were then validated. For ALP < 25 UI/L (model 1), the area under curve (AUC) and the 95% confidence intervals (CI) was 0.68 (95% CI 0.63–0.72) and it improved to 0.87 (95% CI 0.8–0.9), when PEA or PLP threshold levels were added (models 2 and 3), reaching 0.94 (0.91–0.97) when both substrates were included (model 4). The internal validation showed that the addition of serum PLP threshold levels to the model just including ALP improved significantly sensitivity (S) and negative predictive value (NPV) − 100%, respectively- with an accuracy (AC) of 93% in comparison to the inclusion of urinary PEA (S: 71%; NPV 75% and AC: 79%) and similar diagnostic utility measures as those observed in model 3 were detected when both substrates were added.ConclusionsIn this study, we propose a biochemical predictive model based on the threshold levels of the main biochemical markers of HPP (ALP < 25 IU/L and PLP > 180 nmol/L) that when combined, seem to be very useful to identify individuals with ALPL variants.

A Binary Machine Learning Cuckoo Search Algorithm Improved by a Local Search Operator for the Set-Union Knapsack Problem

Optimization techniques, specially metaheuristics, are constantly refined in order to decrease execution times, increase the quality of solutions, and address larger target cases. Hybridizing techniques are one of these strategies that are particularly noteworthy due to the breadth of applications. In this article, a hybrid algorithm is proposed that integrates the k-means algorithm to generate a binary version of the cuckoo search technique, and this is strengthened by a local search operator. The binary cuckoo search algorithm is applied to the NP-hard Set-Union Knapsack Problem. This problem has recently attracted great attention from the operational research community due to the breadth of its applications and the difficulty it presents in solving medium and large instances. Numerical experiments were conducted to gain insight into the contribution of the final results of the k-means technique and the local search operator. Furthermore, a comparison to state-of-the-art algorithms is made. The results demonstrate that the hybrid algorithm consistently produces superior results in the majority of the analyzed medium instances, and its performance is competitive, but degrades in large instances.

Occluded Image Function: A Novel Measure for Evaluating Machine Learning Classifiers for Biometrics

A novel method is presented for evaluating the efficacy of object recognition algorithms on occluded images, called the occluded image function (OIF). The OIF describes system behavior in occluded environments and thus gives qualitative insight into their mechanisms; derivative metrics from OIF can also be used to quantitatively compare classifiers. To showcase the utility of the OIF, an experiment is performed by obstructing optical gait images from two biped robot models and using four binary machine learning classifiers to distinguish between them. The OIF diagrams are created from each experiment, and the resulting insights about the classifiers are discussed. Using the OIF, it is shown that the primitive classifiers can sometimes perform better under occlusion conditions, possibly due to pre-filtering of gait data by uniform occlusions. This result serves to demonstrate that the OIF is a useful tool for classifier evaluation.

EMCNet: Automated COVID-19 diagnosis from X-ray images using convolutional neural network and ensemble of machine learning classifiers.

Recently, coronavirus disease (COVID-19) has caused a serious effect on the healthcare system and the overall global economy. Doctors, researchers, and experts are focusing on alternative ways for the rapid detection of COVID-19, such as the development of automatic COVID-19 detection systems. In this paper, an automated detection scheme named EMCNet was proposed to identify COVID-19 patients by evaluating chest X-ray images. A convolutional neural network was developed focusing on the simplicity of the model to extract deep and high-level features from X-ray images of patients infected with COVID-19. With the extracted features, binary machine learning classifiers (random forest, support vector machine, decision tree, and AdaBoost) were developed for the detection of COVID-19. Finally, these classifiers’ outputs were combined to develop an ensemble of classifiers, which ensures better results for the dataset of various sizes and resolutions. In comparison with other recent deep learning-based systems, EMCNet showed better performance with 98.91% accuracy, 100% precision, 97.82% recall, and 98.89% F1-score. The system could maintain its great importance on the automatic detection of COVID-19 through instant detection and low false negative rate.