Machine Classifier Research Articles

Triple-0: Zero-shot denoising and dereverberation on an end-to-end frozen anechoic speech separation network.

Speech enhancement is crucial both for human and machine listening applications. Over the last decade, the use of deep learning for speech enhancement has resulted in tremendous improvement over the classical signal processing and machine learning methods. However, training a deep neural network is not only time-consuming; it also requires extensive computational resources and a large training dataset. Transfer learning, i.e. using a pretrained network for a new task, comes to the rescue by reducing the amount of training time, computational resources, and the required dataset, but the network still needs to be fine-tuned for the new task. This paper presents a novel method of speech denoising and dereverberation (SD&D) on an end-to-end frozen binaural anechoic speech separation network. The frozen network requires neither any architectural change nor any fine-tuning for the new task, as is usually required for transfer learning. The interaural cues of a source placed inside noisy and echoic surroundings are given as input to this pretrained network to extract the target speech from noise and reverberation. Although the pretrained model used in this paper has never seen noisy reverberant conditions during its training, it performs satisfactorily for zero-shot testing (ZST) under these conditions. It is because the pretrained model used here has been trained on the direct-path interaural cues of an active source and so it can recognize them even in the presence of echoes and noise. ZST on the same dataset on which the pretrained network was trained (homo-corpus) for the unseen class of interference, has shown considerable improvement over the weighted prediction error (WPE) algorithm in terms of four objective speech quality and intelligibility metrics. Also, the proposed model offers similar performance provided by a deep learning SD&D algorithm for this dataset under varying conditions of noise and reverberations. Similarly, ZST on a different dataset has provided an improvement in intelligibility and almost equivalent quality as provided by the WPE algorithm.

Using machine learning to identify environmental factors that collectively determine microbial community structure of activated sludge

Activated sludge (AS) microbial communities are influenced by various environmental variables. However, a comprehensive analysis of how these variables jointly and nonlinearly shape the AS microbial community remains challenging. In this study, we employed advanced machine learning techniques to elucidate the collective effects of environmental variables on the structure and function of AS microbial communities. Applying Dirichlet multinomial mixtures analysis to 311 global AS samples, we identified four distinct microbial community types (AS-types), each characterized by unique microbial compositions and metabolic profiles. We used 14 classical linear and nonlinear machine learning methods to select a baseline model. The extremely randomized trees demonstrated optimal performance in learning the relationship between environmental factors and AS types (with an accuracy of 71.43%). Feature selection identified critical environmental factors and their importance rankings, including latitude (Lat), longitude (Long), precipitation during sampling (Precip), solids retention time (SRT), effluent total nitrogen (Effluent TN), average temperature during sampling month (Avg Temp), mixed liquor temperature (Mixed Temp), influent biochemical oxygen demand (Influent BOD), and annual precipitation (Annual Precip). Significantly, Lat, Long, Precip, Avg Temp, and Annual Precip, influenced metabolic variations among AS types. These findings emphasize the pivotal role of environmental variables in shaping microbial community structures and enhancing metabolic pathways within activated sludge. Our study encourages the application of machine learning techniques to design artificial activated sludge microbial communities for specific environmental purposes.

SDR implementation of wideband spectrum sensing using machine learning

SummaryNew cognitive radio (CR) systems require high throughput and bandwidth. Hence, CR users need to detect wide frequency bands of the radio spectrum to exploit unused frequency channels. This paper proposes a new wideband spectrum sensing (WBSS) detection approach based on machine learning (ML) for scanning subchannels. The originality of the proposed approach is to detect spectrum opportunities using a narrowband spectrum sensing (NBSS) method‐based support vector machine (SVM) classification and two features: energy and goodness of fit (GoF). The simulation results show that the proposed WBSS approach‐based ML presents a higher probability of detection than the WBSS approach‐based conventional detectors, even at low signal‐to‐noise ratio (SNR). Finally, the software defined radio (SDR) implementation validates the proposed WBSS approach for real detection scenarios.

Large Language Model-Based Natural Language Encoding Could Be All You Need for Drug Biomedical Association Prediction.

Analyzing drug-related interactions in the field of biomedicine has been a critical aspect of drug discovery and development. While various artificial intelligence (AI)-based tools have been proposed to analyze drug biomedical associations (DBAs), their feature encoding did not adequately account for crucial biomedical functions and semantic concepts, thereby still hindering their progress. Since the advent of ChatGPT by OpenAI in 2022, large language models (LLMs) have demonstrated rapid growth and significant success across various applications. Herein, LEDAP was introduced, which uniquely leveraged LLM-based biotext feature encoding for predicting drug-disease associations, drug-drug interactions, and drug-side effect associations. Benefiting from the large-scale knowledgebase pre-training, LLMs had great potential in drug development analysis owing to their holistic understanding of natural language and human topics. LEDAP illustrated its notable competitiveness in comparison with other popular DBA analysis tools. Specifically, even in simple conjunction with classical machine learning methods, LLM-based feature representations consistently enabled satisfactory performance across diverse DBA tasks like binary classification, multiclass classification, and regression. Our findings underpinned the considerable potential of LLMs in drug development research, indicating a catalyst for further progress in related fields.

Investigating the capability of recurrent neural networks for identifying text-based cyberbullying incidents on social platforms

As cyberbullying escalates worldwide, the emotional toll on those targeted is profound, demanding attention. This question allows the framing of the detection of cyberbullying as a categorization puzzle, leveraging a spectrum of detection strategies from classical machine learning to advanced deep learning techniques via Natural Language Processing (NLP). It zeroes down to that capability which characterizes RNNs in parsing sequential data and interpreting the contextual nuances. This hence underscores their effectiveness and accuracy in the flagged detection of cyberbullying content. Directly compared to classical algorithms, it shows the best performance of RNN regarding accuracy, speed, and universality over different languages. The overall result of this research does, therefore, affirm the very strong promise and effectiveness of RNN frameworks toward the discrimination of cyberbullying across the various linguistic online environments, setting a firm ground for further development in cybersecurity.

Near-Term Quantum Classification Algorithms Applied to Antimalarial Drug Discovery.

Computational approaches are widely applied in drug discovery to explore properties related to bioactivity, physiochemistry, and toxicology. Over at least the last 20 years, the exploitation of machine learning on molecular data sets has been used to understand the structure-activity relationships that exist between biomolecules and druggable targets. More recently, these methods have also seen application for phenotypic screening data for neglected diseases such as tuberculosis and malaria. Herein, we apply machine learning to build quantum Quantitative Structure Activity Relationship models from antimalarial data sets. There is a continual need for new antimalarials to address drug resistance, and the readily available in vitro data sets could be utilized with newer machine learning approaches as these develop. Furthermore, quantum machine learning is a relatively new method that uses a quantum computer to perform the calculations. First, we present a classical-quantum hybrid computational approach by building a Latent Bernoulli Autoencoder machine learning model for compressing bit-vector descriptors to a size that can be adapted to quantum computers for classification tasks with limited loss of embedded information. Second, we apply our method for feature map compression to quantum classification algorithms, including a completely novel machine learning algorithm with no analogy in classical computers: the Quantum Fourier Transform Classifier. We apply both these approaches to build quantum machine learning models for small-molecule antimalarials with quantum simulation software and then benchmark these quantum models against classical machine learning approaches. While there are many challenges currently facing the development of reliable quantum computers, our results demonstrate that there is potential for the use of this technology in the field of drug discovery.

Artificial intelligence and machine learning in optics: tutorial

Across the spectrum of scientific inquiry and practical applications, the emergence of artificial intelligence (AI) and machine learning (ML) has comprehensively revolutionized problem-solving methodologies. This tutorial explores key aspects of AI/ML and their remarkable role in augmenting the capabilities of optics and photonics technologies. Beginning with fundamental definitions and paradigms, the tutorial progresses to classical machine learning algorithms, with examples employing support vector machines and random forests. Extensive discussion of deep learning encompasses the backpropagation algorithm and artificial neural networks, with examples demonstrating the applications of dense and convolutional neural networks. Data augmentation and transfer learning are examined next as effective strategies for handling scenarios with limited datasets. Finally, the necessity of alleviating the burden of data collection and labeling is discussed, motivating the investigation of unsupervised and semi-supervised learning strategies as well as the utilization of reinforcement learning. By providing a structured exploration of AI/ML techniques, this tutorial equips researchers with the essential tools to begin leveraging AI’s transformative potential within the expansive realm of optics and photonics.

Development and validation of a deep learning-based model for diabetes diagnosis

Diabetes is a systemic disease caused by hyperglycemia, and the number of people with diabetes worldwide may reach 1.31 billion by 2050. The traditional diagnosis of early diabetes is difficult and inaccurate. Computer-aided automatic method has been widely applied in diabetes diagnosis at early stage. In order to realize automatic diagnosis of diabetes, this study proposed a deep neural network-based model for diabetes diagnosis, and its performance was compared with those of other classical machine leaning models. After data cleaning, Synthetic Minority Over-sampling Technique (SMOTE) sampling and feature selecting of the diabetes dataset, multiple models were applied to the features such as Body Mass Index (BMI), Age and other features for prediction. The results showed that Deep Neural Network (DNN) is the best diagnostic solution for diabetes with excellent performance with an accuracy of 99.5%. More specific features will be considered to ensure the accuracy and credibility of clinical diagnosis of diabetes.

Diagnosis of pregnancy disorder in the first‐trimester patient plasma with Raman spectroscopy and protein analysis

AbstractGestational diabetes mellitus (GDM) is a pregnancy disorder associated with short‐ and long‐term adverse outcomes in both mothers and infants. The current clinical test of blood glucose levels late in the second trimester is inadequate for early detection of GDM. Here we show the utility of Raman spectroscopy (RS) for rapid and highly sensitive maternal metabolome screening for GDM in the first trimester. Key metabolites, including phospholipids, carbohydrates, and major amino acids, were identified with RS and validated with mass spectrometry, enabling insights into associated metabolic pathway enrichment. Using classical machine learning (ML) approaches, we showed the performance of the RS metabolic model (cross‐validation AUC 0.97) surpassed that achieved with patients' clinical data alone (cross‐validation AUC 0.59) or prior studies with single biomarkers. Further, we analyzed novel proteins and identified fetuin‐A as a promising candidate for early GDM prediction. A correlation analysis showed a moderate to strong correlation between multiple metabolites and proteins, suggesting a combined protein‐metabolic analysis integrated with ML would enable a powerful screening platform for first trimester diagnosis. Our study underscores RS metabolic profiling as a cost‐effective tool that can be integrated into the current clinical workflow for accurate risk stratification of GDM and to improve both maternal and neonatal outcomes.

Investigating the capability of recurrent neural networks for identifying text-based cyberbullying incidents on social platforms

As cyberbullying escalates worldwide, the emotional toll on those targeted is profound, demanding attention. This question allows the framing of the detection of cyberbullying as a categorization puzzle, leveraging a spectrum of detection strategies from classical machine learning to advanced deep learning techniques via Natural Language Processing (NLP). It zeroes down to that capability which characterizes RNNs in parsing sequential data and interpreting the contextual nuances. This hence underscores their effectiveness and accuracy in the flagged detection of cyberbullying content. Directly compared to classical algorithms, it shows the best performance of RNN regarding accuracy, speed, and universality over different languages. The overall result of this research does, therefore, affirm the very strong promise and effectiveness of RNN frameworks toward the discrimination of cyberbullying across the various linguistic online environments, setting a firm ground for further development in cybersecurity.

A hyperparameter study for quantum kernel methods

Quantum kernel methods are a promising method in quantum machine learning thanks to the guarantees connected to them. Their accessibility for analytic considerations also opens up the possibility of prescreening datasets based on their potential for a quantum advantage. To do so, earlier works developed the geometric difference, which can be understood as a closeness measure between two kernel-based machine learning approaches, most importantly between a quantum kernel and a classical kernel. This metric links the quantum and classical model complexities, and it was developed to bound generalization error. Therefore, it raises the question of how this metric behaves in an empirical setting. In this work, we investigate the effects of hyperparameter choice on the model performance and the generalization gap between classical and quantum kernels. The importance of hyperparameters is well known also for classical machine learning. Of special interest are hyperparameters associated with the quantum Hamiltonian evolution feature map, as well as the number of qubits to trace out before computing a projected quantum kernel. We conduct a thorough investigation of the hyperparameters across 11 datasets, and we identify certain aspects that can be exploited. Analyzing the effects of certain hyperparameter settings on the empirical performance, as measured by cross validation accuracy, and generalization ability, as measured by geometric difference described above, brings us one step closer to understanding the potential of quantum kernel methods on classical datasets.

Analysis of support vector machine and random forest models for classification of the impact of technostress in covid and post-covid era

This study addresses the growing concern of technostress, a condition caused by the overwhelming use of digital technologies, exacerbated by the COVID-19 pandemic. The researchers developed a predictive model using machine learning algorithms, Random Forest (RF) and Support Vector Machine (SVM), to assess and manage technostress levels. The model considers factors such as age, gender, technology usage hours, and technological experiences to classify stress levels into high, moderate, and low categories. The study collected data through a questionnaire administered to knowledgeable respondents, using a non-probabilistic sampling approach. The results showed that both RF and SVM algorithms achieved high accuracy in classifying technostress, with SVM performing slightly better (94.5% vs 84.50%). The model’s effectiveness in predicting stress levels for users with varying degrees of stress is a significant contribution to the field. The research also developed an interactive user interface to facilitate user engagement with the model, promoting stress management and well-being in a technology-driven society. The study’s findings provide valuable insights into the challenges posed by technostress and offer a solution for mitigating its effects. The use of machine learning algorithms to classify gender based on the dataset demonstrates the model’s potential applications in various areas. Overall, this study demonstrates the importance of addressing technostress in the digital age and provides a valuable tool for managing stress levels. The development of predictive models like this one can help individuals and organizations mitigate the negative impacts of technostress, promoting a healthier and more sustainable relationship with technology.

Rapid and non-destructive identification of Panax ginseng origins using hyperspectral imaging, visible light imaging, and X-ray imaging combined with multi-source data fusion strategies

The geographical origin of Panax ginseng significantly influences its nutritional value and chemical composition, which in turn affects its market price. Traditional methods for analyzing these differences are often time-consuming and require substantial quantities of reagents, rendering them inefficient. Therefore, hyperspectral imaging (HSI) in conjunction with X-ray technology were used for the swift and non-destructive traceability of Panax ginseng origin. Initially, outlier samples were effectively rejected by employing a combined isolated forest algorithm and density peak clustering (DPC) algorithm. Subsequently, random forest (RF) and support vector machine (SVM) classification models were constructed using hyperspectral spectral data. These models were further optimized through the application of 72 preprocessing methods and their combinations. Additionally, to enhance the model’s performance, four variable screening algorithms were employed: SelectKBest, genetic algorithm (GA), least absolute shrinkage and selection operator (LASSO), and permutation feature importance (PFI). The optimized model, utilizing second derivative, auto scaling, permutation feature importance, and support vector machine (2nd Der-AS-PFI-SVM), achieved a prediction accuracy of 93.4 %, a Kappa value of 0.876, a Brier score of 0.030, an F1 score of 0.932, and an AUC of 0.994 on an independent prediction set. Moreover, the image data (including color information and texture information) extracted from color and X-ray images were used to construct classification models and evaluate their performance. Among them, the SVM model constructed using texture information from X -ray images performed the best, and it achieved a prediction accuracy of 63.0 % on the validation set, with a Brier score of 0.181, an F1 score of 0.518, and an AUC of 0.553. By implementing mid-level fusion and high-level data fusion based on the Stacking strategy, it was found that the model employing a high-level fusion of hyperspectral spectral information and X-ray images texture information significantly outperformed the model using only hyperspectral spectral information. This advanced model attained a prediction accuracy of 95.2 %, a Kappa value of 0.912, a Brier score of 0.027, an F1 score of 0.952, and an AUC of 0.997 on the independent prediction set. In summary, this study not only provides a novel technical path for fast and non-destructive traceability of Panax ginseng origin, but also demonstrates the great potential of the combined application of HSI and X-ray technology in the field of traceability of both medicinal and food products.

An Analysis of Traditional Methods and Deep Learning Methods in SSVEP-Based BCI: A Survey

The brain–computer interface (BCI) is a direct communication channel between humans and machines that relies on the central nervous system. Neuroelectric signals are collected by placing electrodes, and after feature sampling and classification, they are converted into control signals to control external mechanical devices. BCIs based on steady-state visual evoked potential (SSVEP) have the advantages of high classification accuracy, fast information conduction rate, and relatively strong anti-interference ability, so they have been widely noticed and discussed. From k-nearest neighbor (KNN), multilayer perceptron (MLP), and support vector machine (SVM) classification algorithms to the current deep learning classification algorithms based on neural networks, a wide variety of discussions and analyses have been conducted by numerous researchers. This article summarizes more than 60 SSVEP- and BCI-related articles published between 2015 and 2023, and provides an in-depth research and analysis of SSVEP-BCI. The survey in this article can save a lot of time for scholars in understanding the progress of SSVEP-BCI research and deep learning, and it is an important guide for designing and selecting SSVEP-BCI classification algorithms.

Probabilistic Confusion Matrix: A Novel Method for Machine Learning Algorithm Generalized Performance Analysis

The paper addresses the issue of classification machine learning algorithm performance based on a novel probabilistic confusion matrix concept. The paper develops a theoretical framework which associates the proposed confusion matrix and the resulting performance metrics with the regular confusion matrix. The theoretical results are verified based on a wide variety of real-world classification problems and state-of-the-art machine learning algorithms. Based on the properties of the probabilistic confusion matrix, the paper then highlights the benefits of using the proposed concept both during the training phase and the application phase of a classification machine learning algorithm.

Machine learning driven performance enhancement of perovskite solar cells with CNT as both hole transport layer and back contact

We propose a machine learning-based approach to enhance the performance of perovskite solar cells (PSCs) using carbon nanotubes as both hole transport layer (HTL) and back contact. Carbon-based PSCs offer low fabrication costs, long-term mechanical stability, high charge transport, and broad wavelength transparency. Our study estimates and enhances power conversion efficiency (PCE) of carbon-based PSCs through machine learning. PSC with single-walled carbon nanotube (SWCNT)-based HTL and back contact is investigated in simulations using SCAPS-1D and achieved 14.24% PCE, closely aligning with experimental data. Additionally, we explore the impact of various easily tunable fabrication parameters, such as the band gap and electron affinity of SWCNT in HTL, HTL thickness, active layer thickness, electron transport layer (ETL) thickness, HTL dopant concentration, absorber defect density, and ETL dopant concentration, on PCE. We generate a dataset of 20,480 samples to train classical and neural network-based machine learning models to predict performance parameters of PSCs, including short-circuit current density (JSC), open-circuit voltage (VOC), fill factor (FF), and PCE (η). The current density vs. voltage (J-V) characteristics curve is well-fitted by the predicted JSC, VOC, and FF, as evidenced by an average root mean squared error (RMSE) of 0.03 and a goodness of fit (R2) value of 0.99. The fully connected artificial neural network performs best, with RMSE of 0.05% and R2 of 0.99998 for predicting PCE. Genetic algorithm is used to optimize PCE, yielding a maximum PCE of 20.92%. Furthermore, an analysis of feature dimension and dataset size on prediction performance guides future modeling approaches.

A multi-information fusion anomaly detection model based on convolutional neural networks and AutoEncoder

Network traffic anomaly detection, as an effective analysis method for network security, can identify differentiated traffic information and provide secure operation in complex and changing network environments. To avoid information loss caused when handling traffic data while improving the detection performance of traffic feature information, this paper proposes a multi-information fusion model based on a convolutional neural network and AutoEncoder. The model uses a convolutional neural network to extract features directly from the raw traffic data, and a AutoEncoder to encode the statistical features extracted from the raw traffic data, which are used to supplement the information loss due to cropping. These two features are combined to form a new integrated feature for network traffic, which has the load information from the original traffic data and the global information of the original traffic data obtained from the statistical features, thus providing a complete representation of the information contained in the network traffic and improving the detection performance of the model. The experiments show that the classification accuracy of network traffic anomaly detection using this model outperforms that of classical machine learning methods.

Predicting Recurrent Deficiency and Suboptimal Monitoring of Thiamin Deficiency in Patients with Metabolic and Bariatric Surgery.

Vitamin B1 (thiamine) deficiency (TD) after metabolic and bariatric surgery (MBS) is often insidious and, if unrecognized, can lead to irreversible damage or death. As TD symptoms are vague and overlap with other disorders, we aim to identify predictors of recurrent TD and failure to collect B1 labs. We analyzed a large sample of data from patients with MBS (n = 878) to identify potential predictors of TD risk. We modeled recurrent TD and failure to collect B1 labs using classical statistical and machine learning (ML) techniques. We identified clusters of labs associated with increased risk of recurrent TD: micronutrient deficiencies, abnormal blood indices, malnutrition, and fluctuating electrolyte levels (aIRR range: 1.62-4.68). Additionally, demographic variables associated with lower socioeconomic status were predictive of recurrent TD. ML models predicting characteristics associated with failure to collect B1 labs achieved 75-81% accuracy, indicating that clinicians may fail to match symptoms with the underlying condition. Our analysis suggests that both clinical and social factors can increase the risk of life-threatening TD episodes in some MBS patients. Identifying these indicators can help with diagnosis and treatment.

Analysis of Devanur and Manamedu Ophiolite Complexes in SGT, India: A detailed examination employing remote sensing techniques and Laboratory Spectral Signature investigations

This study employs advanced satellite imagery from ASTER and Sentinel-2A to conduct detailed lithological mapping of the Devanur and Manamedu ophiolite complexes in the southern Central Shear Zone (CSZ). The primary focus is on the Manamedu Ophiolite Complex (MOC) and the Devanur Ophiolitic Complex (DOC). Image enhancement techniques such as Color composites, Principal Component Analysis (PCA), and Minimum Noise Fraction (MNF) were utilized to differentiate various rock types. RGB band combinations derived from PCA and MNF outputs demonstrated effective discrimination of rock units. Spectral Angle Mapper (SAM) and Support Vector Machine (SVM) classification methods were employed on ASTER and Sentinel-2A images, yielding classified lithologies that closely matched existing maps from the Geological Survey of India (GSI) and other studies, validating the accuracy of the findings. Additionally, Laboratory Spectral Signature Studies were conducted on 10 rock samples using an ASD FieldSpec Pro® spectroradiometer, providing reflectance spectra from 350 nm to 2500 nm. These spectra, particularly the continuum-removed reflectance, revealed diagnostic absorption features that were corroborated by geochemical analyses. A detailed analysis investigated how elemental compositions and key minerals influenced absorption bands. Major oxide geochemical compositions of DOC and MOC samples were identified using XRF methods. The aim of this research is to characterize DOC and MOC through remote sensing and spectral signature analysis. Sentinel-2A data proved more effective in lithological discrimination compared to ASTER, with spectral signatures indicating the presence of iron (Fe) and magnesium (Mg) contents. Notably, SVM classification of Sentinel-2A MNF + DEM data achieved an overall accuracy of more than 90% when compared with field investigations. This study underscores the efficacy of processing VNIR and SWIR bands from ASTER and Sentinel-2A satellite imagery alongside DEM data and ground surveys for mapping mafic-ultramafic rocks in the DOC and MOC regions of the CSZ.

Epilepsy lesion localization method based on brain function network.

In the past, the localization of seizure onset zone (SOZ) primarily relied on traditional EEG signal analysis methods. However, due to their limited spatial and temporal resolution, accurately pinpointing neural activity was challenging, thereby restricting their clinical applicability. Compared with traditional EEG signals, SEEG signals have superior spatial and temporal resolution, and can more accurately record neural activity near epileptic foci, making them better suited for studying SOZ. In addition, the traditional EEG signal analysis methods still have limitations, mainly focusing on the analysis of local signal features, while ignoring the complexity and interconnection of the overall brain network. How to more accurately locate SOZ is still not well resolved. The purpose of this study is to develop an effective positioning method for more accurate positioning. To overcome these limitations, this study proposed a model integrating brain functional network analysis with nonlinear dynamics. We utilized weighted phase lag index (WPLI) to construct brain functional network, epilepic network connectivity strength (ENCS) as the feature, and introduced persistence entropy (PE) for feature fusion, subsequently employing support vector machine (SVM) classification. The proposed method was verified on the HUP-iEEG dataset, our solution identified the SOZ with 0.9440 accuracy, 0.9848 precision, 0.8974 recall rate, 0.9340 F1 score and 0.9697 area under the ROC curve across patients, which outperforms the existing approaches. It exhibits a 2.30 percentage point enhancement in localisation accuracy along with a 2.97 percentage points in AUC compared to others. Our method consider the interactions between nodes in brain network connections, as well as the inherent nonlinear and non-stationary properties of neural signals, to be more robust.