Multiple Classification Research Articles

Pseudo-Bag Mixup Augmentation for Multiple Instance Learning-Based Whole Slide Image Classification.

Given the special situation of modeling gigapixel images, multiple instance learning (MIL) has become one of the most important frameworks for Whole Slide Image (WSI) classification. In current practice, most MIL networks often face two unavoidable problems in training: i) insufficient WSI data and ii) the sample memorization inclination inherent in neural networks. These problems may hinder MIL models from adequate and efficient training, suppressing the continuous performance promotion of classification models on WSIs. Inspired by the basic idea of Mixup, this paper proposes a new Pseudo-bag Mixup (PseMix) data augmentation scheme to improve the training of MIL models. This scheme generalizes the Mixup strategy for general images to special WSIs via pseudo-bags so as to be applied in MIL-based WSI classification. Cooperated by pseudo-bags, our PseMix fulfills the critical size alignment and semantic alignment in Mixup strategy. Moreover, it is designed as an efficient and decoupled method, neither involving time-consuming operations nor relying on MIL model predictions. Comparative experiments and ablation studies are specially designed to evaluate the effectiveness and advantages of our PseMix. Experimental results show that PseMix could often assist state-of-the-art MIL networks to refresh their classification performance on WSIs. Besides, it could also boost the generalization performance of MIL models in special test scenarios, and promote their robustness to patch occlusion and label noise. Our source code is available at https://github.com/liupei101/PseMix.

CRISHI :A Crop Recommendation System

Abstract: The Crop Recommendation System proposes a mixed model to suggest crops for Indian states, considering various factors such as soil pH, moisture, NPK levels, crop data, and temperature. This suggestion model is constructed as a sample model incorporating supervised machine learning algorithms. The technology- based agricultural crop identifier model aims to enhance farmers crop yield by identifying appropriate crops based on their parameters. The efficacy of the recommendation model lies in its ability to recommend the right crop for specific conditions, which in turn contributes to addressing agricultural and farming challenges and boosting the Indian economy by maximizing crop production. The system also includes a grading process to identify crop quality and signalling the presence of low and high-quality crops. The utilization of a set of classifiers facilitates improved decision-making through multiple classifiers. The assessment process includes a request for decisionmaking to determine the outcomes of the classifiers

Speech emotion recognition using a combination of variational mode decomposition and Hilbert transform

The objective of automated speech emotion recognition (SER) is to recognize each emotion of the speech signal uniquely and efficiently using machines like computers, mobile devices, etc. The popularity of SER has been widely acknowledged among researchers due to its extensive applicability in practical contexts. The use of SER has shown to be advantageous in several domains, including medical treatment, enhancement of security systems, surveillance operations, online marketing strategies, online educational platforms, online search engines, personal communication, customer relationship management, reinforcement of machine and human connection, and numerous other areas. Numerous authors have used various techniques, including the combination of multiple features (acoustic, non-acoustic, or both) and classifiers (machine learning, deep learning, or both), in order to enhance the efficacy of emotion categorization. In the present work, our objective is to enhance the performance of emotion classification (PECL) by using a combined approach of the variational mode decomposition (VMD) and Hilbert transform (HT) techniques. Using the VMD method, we decomposed the speech signal frame into many sub-signals, or intrinsic mode functions (IMFs). Then HT is applied to each VMD-based IMF signal to find the mode instantaneous amplitude (MIA) and mode instantaneous frequency (MIF) signal vectors. We extracted proposed features such as HT-based approximate entropy (HTAE), HT-based permutation entropy (HTPE), HT-based increment entropy (HTIE), and HT-based sample entropy (HTSE) using each MIA and MIF signal vector. The combination of the proposed HT-based features is called HT-based entropy (HTE) features. Then, we accessed the PECL using the HTE and MFCC features alone and in conjunction with a deep neural network (DNN) classifier. The experiment's results showed that the combinations of the proposed feature (MFCC + HTE) using a DNN classifier outperformed the individual features and obtained a SER accuracy of 86.92% for the EMOVO dataset and 91.63% for the EMO-DB dataset.

Integrates Differential Gene Expression analysis and deep learning for accurate and robust prostate cancer diagnosis

The challenge of diagnosing complex diseases and increasing human lifespan is a pressing task. Traditional methods, relying on visual characteristics like ultrasound and angiography, often struggle to detect cancer in its early stages, limiting diagnostic accuracy due to the intricate and nonlinear nature of diseases. From the perspective of gene expression, detecting cancer offers a more robust and effective approach due to its ability to directly assess the genetic activity within cells. In this study, we present the development of a prostate cancer feature selection method based on differentially expressed genes (DEGs). Utilizing datasets from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA), we meticulously curated data for both model training and testing, implementing stringent filtering criteria based on p-value and fold change. Our study identifies a panel of 220 genes with substantial potential for prostate cancer detection. We then construct an ANN model for the diagnosis of the disease, whose accuracy is 0.780.01, which is more effective than other models like Ridge Classifiers, Logistic Regression, Naive Bayes Regression and Decision Trees. The average accuracy of these classifiers is 0.730.01. Notably, these genes also demonstrate exceptional performance across other various classifiers, indicating their robustness and effectiveness without dependence on specific models. The credibility is validated by comparison to random genes, and adaptability by using pancreatic cancer data from GEO. The Gene Otology analysis also verifies the feasibility of such method. This panel establishes a solid foundation for advancing clinical diagnostics of prostate cancer. This framework holds potential to significantly transform prostate cancer screening by offering strong resilience and precision across multiple classification methods.

Innovative Decision Fusion for Accurate Crop/Vegetation Classification with Multiple Classifiers and Multisource Remote Sensing Data

Obtaining accurate and real-time spatial distribution information regarding crops is critical for enabling effective smart agricultural management. In this study, innovative decision fusion strategies, including Enhanced Overall Accuracy Index (E-OAI) voting and the Overall Accuracy Index-based Majority Voting (OAI-MV), were introduced to optimize the use of diverse remote sensing data and various classifiers, thereby improving the accuracy of crop/vegetation identification. These strategies were utilized to integrate crop/vegetation classification outcomes from distinct feature sets (including Gaofen-6 reflectance, Sentinel-2 time series of vegetation indices, Sentinel-2 time series of biophysical variables, Sentinel-1 time series of backscatter coefficients, and their combinations) using distinct classifiers (Random Forests (RFs), Support Vector Machines (SVMs), Maximum Likelihood (ML), and U-Net), taking two grain-producing areas (Site #1 and Site #2) in Haixi Prefecture, Qinghai Province, China, as the research area. The results indicate that employing U-Net on feature-combined sets yielded the highest overall accuracy (OA) of 81.23% and 91.49% for Site #1 and Site #2, respectively, in the single classifier experiments. The E-OAI strategy, compared to the original OAI strategy, boosted the OA by 0.17% to 6.28%. Furthermore, the OAI-MV strategy achieved the highest OA of 86.02% and 95.67% for the respective study sites. This study highlights the distinct strengths of various remote sensing features and classifiers in discerning different crop and vegetation types. Additionally, the proposed OAI-MV and E-OAI strategies effectively harness the benefits of diverse classifiers and multisource remote sensing features, significantly enhancing the accuracy of crop/vegetation classification.

Large-scale land use/land cover extraction from Landsat imagery using feature relationships matrix based deep-shallow learning

Deep learning has demonstrated its effectiveness in capturing high-level features, with convolutional neural networks (CNNs) excelling in remote sensing classification. However, CNNs encounter challenges when applied to Landsat images with limited multi-spectral bands, as they struggle to learn stable features from the spectral domain and integrate them with spatial features to enhance accuracy. Additionally, most CNN applications focus on learning features directly from the raw image, making them susceptible to spectral environment changes. To overcome these limitations, this paper introduces a novel approach for large-scale Land Use/Land Cover (LULC) extraction from Landsat OLI images. The proposed classification architecture comprises two modules. The first module utilizes a feature relationships matrix to generate an extent feature map (EFM), and a specifically designed CNN structure learns deep features from the EFM and spatial domain. In the second module, a multiple classifiers system (MCS) is employed to obtain shallow learning features, which are further enhanced by another CNN structure through continued learning. The combined features from these modules contribute to improved classification of remote sensing images. Experimental results demonstrate that our proposed method effectively acquires stable features for training deep learning models with strong generalization ability. It exhibits significant advancements in accuracy improvement and large-scale LULC extraction in the Yangtze River Economic Belt (YREB) in China, surpassing comparative approaches based on deep learning and non-deep learning methods.

Polymorphic Clustering and Approximate Masking Framework for Fine-Grained Insect Image Classification

Insect diversity monitoring is crucial for biological pest control in agriculture and forestry. Modern monitoring of insect species relies heavily on fine-grained image classification models. Fine-grained image classification faces challenges such as small inter-class differences and large intra-class variances, which are even more pronounced in insect scenes where insect species often exhibit significant morphological differences across multiple life stages. To address these challenges, we introduce segmentation and clustering operations into the image classification task and design a novel network model training framework for fine-grained classification of insect images using multi-modality clustering and approximate mask methods, named PCAM-Frame. In the first stage of the framework, we adopt the Polymorphic Clustering Module, and segmentation and clustering operations are employed to distinguish various morphologies of insects at different life stages, allowing the model to differentiate between samples at different life stages during training. The second stage consists of a feature extraction network, called Basenet, which can be any mainstream network that performs well in fine-grained image classification tasks, aiming to provide pre-classification confidence for the next stage. In the third stage, we apply the Approximate Masking Module to mask the common attention regions of the most likely classes and continuously adjust the convergence direction of the model during training using a Deviation Loss function. We apply PCAM-Frame with multiple classification networks as the Basenet in the second stage and conduct extensive experiments on the Insecta dataset of iNaturalist 2017 and IP102 dataset, achieving improvements of 2.2% and 1.4%, respectively. Generalization experiments on other fine-grained image classification datasets such as CUB200-2011 and Stanford Dogs also demonstrate positive effects. These experiments validate the pertinence and effectiveness of our framework PCAM-Frame in fine-grained image classification tasks under complex conditions, particularly in insect scenes.

Ensemble feature ranking approach for software fault prediction

Software fault prediction, which aims to find and fix probable flaws before they appear in real-world settings, is an essential component of software quality assurance. This article provides a thorough analysis of the use of feature ranking algorithms for successful software failure prediction. In order to choose and prioritise the software metrics or qualities most important to fault prediction models, feature ranking approaches are essential. The proposed focus on applying an ensemble feature ranking algorithm to a specific software fault dataset, addressing the challenge posed by the dataset’s high dimensionality. In this extensive study, we examined the effectiveness of multiple machine learning classifiers on six different software projects: jedit, ivy, prop, xerces, tomcat, and poi, utilising feature selection strategies. In order to evaluate classifier performance under two scenarios—one with the top 10 features and another with the top 15 features—our study sought to determine the most relevant features for each project. SVM consistently performed well across the six datasets, achieving noteworthy results like 98.74% accuracy on “jedit” (top 10 features) and 91.88% on “tomcat” (top 10 features). Random Forest achieving 89.20% accuracy on the top 15 features, on “ivy.” In contrast, NB repeatedly recording the lowest accuracy rates, such as 51.58% on “poi” and 50.45% on “xerces” (the top 15 features). These findings highlight SVM and RF as the top performers, whereas NB was consistently the least successful classifier. The findings suggest that the choice of feature ranking algorithm has a substantial impact on the fault prediction models’ predictive accuracy and effectiveness. When using various ranking systems, the research also analyses the trade-offs between computing complexity and forecast accuracy.

Strategic Machine Learning Optimization for Cardiovascular Disease Prediction and High-Risk Patient Identification

Despite medical advancements in recent years, cardiovascular diseases (CVDs) remain a major factor in rising mortality rates, challenging predictions despite extensive expertise. The healthcare sector is poised to benefit significantly from harnessing massive data and the insights we can derive from it, underscoring the importance of integrating machine learning (ML) to improve CVD prevention strategies. In this study, we addressed the major issue of class imbalance in the Behavioral Risk Factor Surveillance System (BRFSS) 2021 heart disease dataset, including personal lifestyle factors, by exploring several resampling techniques, such as the Synthetic Minority Oversampling Technique (SMOTE), Adaptive Synthetic Sampling (ADASYN), SMOTE-Tomek, and SMOTE-Edited Nearest Neighbor (SMOTE-ENN). Subsequently, we trained, tested, and evaluated multiple classifiers, including logistic regression (LR), decision trees (DTs), random forest (RF), gradient boosting (GB), XGBoost (XGB), CatBoost, and artificial neural networks (ANNs), comparing their performance with a primary focus on maximizing sensitivity for CVD risk prediction. Based on our findings, the hybrid resampling techniques outperformed the alternative sampling techniques, and our proposed implementation includes SMOTE-ENN coupled with CatBoost optimized through Optuna, achieving a remarkable 88% rate for recall and 82% for the area under the receiver operating characteristic (ROC) curve (AUC) metric.

Analysis of EEG Signals Between Motor Imaginary Tasks and Rest Condition for Biometric Application

Biometric technology has gained immense popularity as an effective solution for enhancing cybersecurity, specifically in countering financial fraud and security threats. EEG-based authentication is unique among biometric authentication methods due to its unparalleled confidentiality and non-replicability. This study explores the feasibility of using motor imagery tasks and rest conditions for human authentication. Ten physically fit subjects, aged between 20 and 28 years, participated voluntarily in the study. The subjects perform imaginary tasks involving their left and right-hand movement. Each task lasted for two minutes, separated by a one-minute break, and EEG data were collected using the EPOC+ device, which features 14-channel electrodes. The sampling frequency was set at 128 Hz. To extract relevant frequency information, Butterworth bandpass filters were employed to extract the alpha (8-13Hz), beta (14-30Hz), and gamma (30-42Hz) frequency bands. Linear features, such as power spectral density (PSD), were obtained using the Welch Method and the Burg Method, while spectral entropy was used to extract non-linear features. Statistical features mean, median, standard deviation, minimum, and maximum were derived from the PSD, and spectral entropy was used as input for the classifiers. Multiple classifiers, including k-nearest neighbor (KNN), support vector machine (SVM), decision tree and Naive Bayes, were employed for the classification task. The Welch method combined with the support vector machine classifier achieved a higher classification accuracy of 96.83% for the beta waves from channels C3, C4, O1, and O2, corresponding to the frontal and occipital lobes. Interestingly, the rest conditions exhibited a higher classification accuracy of 96.83% compared to the motor-imagery tasks, which achieved 96.04%. The utilization of motor imagery tasks and rest conditions, along with the application of advanced classification techniques, holds promise for the development of robust and reliable biometric systems in cybersecurity.

A Bio-Inspired Visual Perception Transformer for Cross-Domain Semantic Segmentation of High-Resolution Remote Sensing Images

Pixel-level classification of very-high-resolution images is a crucial yet challenging task in remote sensing. While transformers have demonstrated effectiveness in capturing dependencies, their tendency to partition images into patches may restrict their applicability to highly detailed remote sensing images. To extract latent contextual semantic information from high-resolution remote sensing images, we proposed a gaze–saccade transformer (GSV-Trans) with visual perceptual attention. GSV-Trans incorporates a visual perceptual attention (VPA) mechanism that dynamically allocates computational resources based on the semantic complexity of the image. The VPA mechanism includes both gaze attention and eye movement attention, enabling the model to focus on the most critical parts of the image and acquire competitive semantic information. Additionally, to capture contextual semantic information across different levels in the image, we designed an inter-layer short-term visual memory module with bidirectional affinity propagation to guide attention allocation. Furthermore, we introduced a dual-branch pseudo-label module (DBPL) that imposes pixel-level and category-level semantic constraints on both gaze and saccade branches. DBPL encourages the model to extract domain-invariant features and align semantic information across different domains in the feature space. Extensive experiments on multiple pixel-level classification benchmarks confirm the effectiveness and superiority of our method over the state of the art.

Evaluation of Effect of Structural Condition on Surface Deterioration for Flexible Pavements Using Traffic Speed Deflection Device

With the introduction of traffic speed deflection devices (TSDD), collecting network-level surface deflection data becomes more efficient and convenient. State highway agencies (SHAs) are paying more attention to the incorporation of structural information into pavement management systems to support maintenance and rehabilitation (M&R) analysis. For those SHAs who conduct M&R analysis based solely on surface condition data, it is crucial to understand how the inclusion of structural condition data may affect the performance model, which structural indices may be well-suited for their needs, and where these indices may be placed in their current flowchart of M&R analysis. To address these issues, in this study, the effects of eight structural indices obtained from TSDD on surface condition deterioration were investigated using TSDD data and surface condition data collected in Tennessee. The effects of structural indices on surface deterioration were evaluated using multiple regression, classification and regression tree analysis, and 1-year deterioration curve analysis. Surface curvature index, SCI_12, was found to be a significant factor influencing surface deteriorations. Structural indices representing condition of base layers and subgrade were less sensitive to the surface deterioration. To facilitate implementation, a flowchart of how to incorporate structural indices into M&R analysis procedure was proposed.

A novel space–spectrum array tile probability random-forest model enhances LULC mapping accuracy on Google Earth Engine: An experiment in Ordos, China

The rapid renewal of land use and land cover (LULC) maps using remote-sensing technologies constitutes a sine qua non for judicious land resource management at both regional and national scales. Existing research conducted on the Google Earth Engine (GEE) platform has overwhelmingly focused on pixel-based LULC classification techniques, often neglecting the role of spatial context via neighbouring valuable pixel information. Remarkably, little attention has been paid to the amalgamation of 3 × 3 neighbouring pixels into a three-dimensional space–spectrum array that can emulate the functionalities of object-based image analysis. In this study, we developed a novel integrated model consisting of a space–spectrum array (SSA) model based on 3 × 3 neighbouring pixels, a tile model based on random forest, and a multiple probabilistic classification model (SSA-TPRF) on the GEE platform to generate a LULC map with high overall accuracy (OA) for Ordos in 2020. Three bimonthly median value images were synthesised and feature collections, including spectral bands and vegetation indices, were constructed. Five experimental groups (EXP1–EXP5) were used to assess the different model combinations. Subsequent validation procedures employed abundant reference samples and compared the results with those of the three extant LULC mapping products. The results showed that EXP2, which was grounded in the tile-based model, yielded an OA of 87.53%, surpassing that of EXP1 (84.99%), which employed a traditional overall model. Furthermore, EXP3, which integrated the multiple probabilistic classification model with the traditional overall model, exhibited an OA of 85.19%, exceeding that of EXP1. A comparison of the five experimental groups using the four regional spatial subtlety features revealed that the EXP5, employing the SSA-TPRF model, successfully decreased the salt and pepper noise. The OA of six tile sizes ranging from 10 km to 100 km were compared, and the highest OA (88.35%) was achieved at a tile size of 25 km. The resultant LULC map in Ordos, derived from the SSA-TPRF model, showed superior OA compared with the extant LULC products. This study thus contributes to a versatile and scalable model within the GEE framework, offering avenues for facile adaptation and recurrent application across disparate geographical locations and temporal settings. The adaptability of this model is particularly advantageous for developing nations and regions typified by diverse landscapes, thereby catalysing the iterative updating of LULC maps through advanced remote-sensing paradigms.

Assessing the Accuracy of Multiple Classification Algorithms Combining Sentinel-1 and Sentinel-2 for the Citrus Crop Classification and spatialization of the Actual Evapotranspiration Obtained from Flux Tower Eddy Covariance: Case Study of Cap Bon, Tunisia

Assessing the Accuracy of Multiple Classification Algorithms Combining Sentinel-1 and Sentinel-2 for the Citrus Crop Classification and spatialization of the Actual Evapotranspiration Obtained from Flux Tower Eddy Covariance: Case Study of Cap Bon, Tunisia

Local feature‐based video captioning with multiple classifier and CARU‐attention

AbstractVideo captioning aims to identify multiple objects and their behaviours in a video event and generate captions for the current scene. This task aims to generate a detailed description of the current video in real‐time using natural language, which requires deep learning to analyze and determine the relationships between interesting objects in the frame sequence. In practice, existing methods typically involve detecting objects in the frame sequence and then generating captions based on features extracted through object coverage locations. Therefore, the results of caption generation are highly dependent on the performance of object detection and identification. This work proposes an advanced video captioning approach that works in adaptively and effectively addresses the interdependence between event proposals and captions. Additionally, an attention‐based multimodel framework is introduced to capture the main context from the frame and sound in the video scene. Also, an intermediate model is presented to collect the hidden states captured from the input sequence, which performs to extract the main features and implicitly produce multiple event proposals. For caption prediction, the proposed method employs the CARU layer with attention consideration as the primary RNN layer for decoding. Experimental results showed that the proposed work achieves improvements compared to the baseline method and also better performance compared to other state‐of‐the‐art models on the ActivityNet dataset, presenting competitive results in the tasks of video captioning.

Label noise correction for crowdsourcing using dynamic resampling

Crowdsourcing provides a cost-effective labeling solution for the acquisition of labeled training samples for machine learning by employing workers on the Internet. A common approach to improving the label quality is to employ a truth inference method to infer integrated labels for samples from their multiple noisy labels obtained from different crowd workers. Although the quality of integrated labels is significantly improved compared with that of the original noisy ones, it still cannot completely eliminate the noises inevitably existing in the integrated labels. To further improve the label quality, this paper proposes a novel label noise correction method for crowdsourcing based on dynamic resampling (DRNC). DRNC first divides the dataset with inferred labels into a clean set and a noisy set through a filter. According to a certain proportion, the clean set and the noisy set are resampled to train multiple heterogeneous classifiers, which form an ensemble classifier. Then, the dataset is divided by the ensemble classifier into a new sub-noisy set and a sub-clean set. The whole process repeats multiple rounds, generating multiple sub-clean sets. Finally, these sub-clean sets are used to train classifiers, which jointly correct the wrong labels in the dataset by voting. Experimental results on 25 simulated and 4 real-world datasets consistently show that the proposed DRNC averagely improves the quality of labels as well as the quality of learned models in the range of 1 to 10 percentage points, compared with four state-of-the-art crowdsourcing noise correction methods.

Prevalence and determinants of areca nut use among pregnant women in India

ABSTRACT Background Areca nut is the fourth most commonly used substance worldwide after nicotine, alcohol, and caffeine. Women in many countries use it in cultural and religious ceremonies. The use of areca nut is harmful to health; but when women consume it during pregnancy, it exerts several harmful effects on the fetus. This paper examines the prevalence and determinants of areca nut use among pregnant women in India. Methods Data from the Global Adult Tobacco Survey-2, India, are used for analysis (N = 1403 pregnant women). Multiple classification analysis and multivariate logistic regression model were applied to estimate the effect of various determinates. Results About 16% pregnant women used areca nut in any form in India. Areca nut use with (24%) and without (13%) tobacco was higher among scheduled tribes. Region, caste, religion, place of residence, educational attainment, occupation, and economic status are found to be important determinants of areca nut use. Conclusion More specific programs which can effectively reduce the use of areca nut among pregnant women should be adopted. Programs related to prevention of areca nut use should be expanded effectively through health awareness interventions during pregnancy, including media messages, and involvement of frontline health workers, especially focusing on the socio-economically backward groups.

Heterogeneous effects of hospital competition on inpatient quality: an analysis of five common diseases in China

BackgroundMany countries has introduced pro-competition policies in the delivery of healthcare to improve medical quality, including China. With the increasing intensity of competition in China's healthcare market, there are rising concerns among policymakers about the impact of hospital competition on quality. This study investigated heterogeneous effects of hospital competition on inpatient quality.MethodsWe analyzed the inpatient discharge dataset and selected chronic obstructive pulmonary disease (COPD), ischemic stroke, pneumonia, hemorrhagic stroke, and acute myocardial infarction (AMI) as representative diseases. A total of 561,429 patients in Sichuan Province in 2017 and 2019 were included. The outcomes of interest were in-hospital mortality and 30-day unplanned readmissions. The Herfindahl–Hirschman Index was calculated using predicted patient flows to measure hospital competition. To address the spatial correlations of hospitals and the structure of the dataset, the multiple membership multiple classification model was employed for analysis.ResultsAmid intensifying competition in the hospital market, our study discerned no marked statistical variance in the risk of inpatient quality across most diseases examined. Amplified competition exhibited a positive correlation with heightened in-hospital mortality for both COPD and pneumonia patients. Elevated competition escalated the risk of 30-day unplanned readmissions for COPD patients, while inversely affecting the risk for AMI patients.ConclusionsThere is the heterogeneous impact of hospital competition on quality across various diseases in China. Policymakers who intend to leverage hospital competition as a tool to enhance healthcare quality must be cognizant of the possible influences of it.

HUMAN ACTIVITY DETECTION USING MMWAVE SENSOR

Human Activity Recognition (HAR) is becoming increasingly useful for applications such as well-being monitoring and personalizing smart spaces. Traditional methods for HAR often require wearable devices or cameras. The former is not feasible for every environment and the latter has strong privacy concerns. The mmWave radar has been shown to be a promising alternative as it does not imply the same privacy concerns and does not require the users to have wearable devices. In this paper we have used a low- cost mmWave radar to generate micro-Doppler spectrograms to ultimately classify different activities. For this, multiple classifiers and methods of spectrogram filtering have been examined. Finally, a Time-Distributed Convolutional Neural Network in conjunction with a Bi-Directional Long Short-Term Memory has attained an average accuracy of 99.62% on a dataset of 5 activities, involving 2 participants.

Deep learning and genetic algorithm-based ensemble model for feature selection and classification of breast ultrasound images

Feature extraction and selection are important techniques in the classification of medical images. Extraction of key features and selection of relevant features are the preliminary processes that are essential for identifying the shape of an object or diagnosis of a tumor in images. In this study, we conduct a thorough comparison of deep neural networks' performance. The comparison of the networks infers MobileNet as optimal for feature extraction from medical images, where it has minimal parameters and high validation accuracy. For feature selection, we employ the Genetic Algorithm (GA) because of its proficiency in handling high-dimensional and complex feature space. GA's iterative process aligns well with the unique characteristics of breast ultrasound (BUS) images, which enhances its efficacy in selecting salient features of BUS images. An ensemble model, capitalizing on the collective decision-making capabilities of multiple classifiers based on a weighted voting scheme for classification, is proposed. Empirical evaluations are conducted using two publicly available BUS image datasets, BUSI and UDIAT. The proposed model demonstrates a notable improvement of approximately 4% and 9% in accuracy for the BUSI and UDIAT, respectively. The improved diagnostic accuracy in breast abnormality identification allows for early abnormality diagnosis. This improves treatment outcomes for breast cancer patients and highlights the practical value of the proposed method for improving BUS image categorization.