Classification Applications Research Articles

SimCDL: A Simple Framework for Contrastive Dictionary Learning

In this paper, we propose a novel approach to the dictionary learning (DL) initialization problem, leveraging the SimCLR framework from deep learning in a self-supervised manner. Dictionary learning seeks to represent signals as sparse combinations of dictionary atoms, but effective initialization remains challenging. By applying contrastive learning, we encourage similar representations for augmented versions of the same sample while distinguishing between different samples. This results in a more diverse and incoherent set of atoms, which enhances the performance of DL applications in classification and anomaly detection tasks. Our experiments across several benchmark datasets demonstrate the effectiveness of our method for improving dictionary learning initialization and its subsequent impact on performance in various applications.

Edge Computing for AI-Based Brain MRI Applications: A Critical Evaluation of Real-Time Classification and Segmentation

Medical imaging plays a pivotal role in diagnostic medicine with technologies like Magnetic Resonance Imagining (MRI), Computed Tomography (CT), Positron Emission Tomography (PET), and ultrasound scans being widely used to assist radiologists and medical experts in reaching concrete diagnosis. Given the recent massive uplift in the storage and processing capabilities of computers, and the publicly available big data, Artificial Intelligence (AI) has also started contributing to improving diagnostic radiology. Edge computing devices and handheld gadgets can serve as useful tools to process medical data in remote areas with limited network and computational resources. In this research, the capabilities of multiple platforms are evaluated for the real-time deployment of diagnostic tools. MRI classification and segmentation applications developed in previous studies are used for testing the performance using different hardware and software configurations. Cost–benefit analysis is carried out using a workstation with a NVIDIA Graphics Processing Unit (GPU), Jetson Xavier NX, Raspberry Pi 4B, and Android phone, using MATLAB, Python, and Android Studio. The mean computational times for the classification app on the PC, Jetson Xavier NX, and Raspberry Pi are 1.2074, 3.7627, and 3.4747 s, respectively. On the low-cost Android phone, this time is observed to be 0.1068 s using the Dynamic Range Quantized TFLite version of the baseline model, with slight degradation in accuracy. For the segmentation app, the times are 1.8241, 5.2641, 6.2162, and 3.2023 s, respectively, when using JPEG inputs. The Jetson Xavier NX and Android phone stand out as the best platforms due to their compact size, fast inference times, and affordability.

FSH3D: 3D Representation via Fibonacci Spherical Harmonics

AbstractSpherical harmonics are a favorable technique for 3D representation, employing a frequency‐based approach through the spherical harmonic transform (SHT). Typically, SHT is performed using equiangular sampling grids. However, these grids are non‐uniform on spherical surfaces and exhibit local anisotropy, a common limitation in existing spherical harmonic decomposition methods. This paper proposes a 3D representation method using Fibonacci Spherical Harmonics (FSH3D). We introduce a spherical Fibonacci grid (SFG), which is more uniform than equiangular grids for SHT in the frequency domain. Our method employs analytical weights for SHT on SFG, effectively assigning sampling errors to spherical harmonic degrees higher than the recovered band‐limited function. This provides a novel solution for spherical harmonic transformation on non‐equiangular grids. The key advantages of our FSH3D method include: 1) With the same number of sampling points, SFG captures more features without bias compared to equiangular grids; 2) The root mean square error of 32‐degree spherical harmonic coefficients is reduced by approximately 34.6% for SFG compared to equiangular grids; and 3) FSH3D offers more stable frequency domain representations, especially for rotating functions. FSH3D enhances the stability of frequency domain representations under rotational transformations. Its application in 3D shape reconstruction and 3D shape classification results in more accurate and robust representations. Our code is publicly available at https://github.com/Miraclelzk/Fibonacci-Spherical-Harmonics.

A Novel Autoencoder-Integrated Clustering Methodology for Inventory Classification: A Real Case Study for White Goods Industry

This article presents an inventory classification method that provides more accurate results in the white goods factory, which will contribute to sustainability, sustainability economics, and supply chain management targets. A novel inventory classification application is presented with real-world data. Two different datasets are used, and these datasets are compared to each other. These larger dataset is Stock Keeping Unit (SKU)-based (6.032 SKUs), and the smaller one is product-group-based (270 product groups). In the first phase, Artificial Intelligence (AI) clustering methods that have not been used in the field of inventory classification, to our knowledge, are applied to these datasets; the results are obtained and compared using K-Means, Gaussian mixture, agglomerative clustering, and spectral clustering methods. In the second stage, an autoencoder is separately hybridized with the AI clustering methods to develop a novel approach to inventory classification. Fuzzy C-Means (FCM) is used in the third step to classify inventories. At the end of the study, these nine different methodologies (“K-Means, Gaussian mixture, agglomerative clustering, spectral clustering” with and without the autoencoder and Fuzzy C-Means) are compared using two different datasets. It is shown that the proposed new hybrid method gives much better results than classical AI methods.

Towards Self-Conscious AI Using Deep ImageNet Models: Application for Blood Cell Classification

The exceptional performance of ImageNet competition winners in image classification has led AI researchers to repurpose these models for a whole range of tasks using transfer learning (TL). TL has been hailed for boosting performance, shortening learning time and reducing computational effort. Despite these benefits, issues such as data sparsity and the misrepresentation of classes can diminish these gains, occasionally leading to misleading TL accuracy scores. This research explores the innovative concept of endowing ImageNet models with a self-awareness that enables them to recognize their own accumulated knowledge and experience. Such self-awareness is expected to improve their adaptability in various domains. We conduct a case study using two different datasets, PBC and BCCD, which focus on blood cell classification. The PBC dataset provides high-resolution images with abundant data, while the BCCD dataset is hindered by limited data and inferior image quality. To compensate for these discrepancies, we use data augmentation for BCCD and undersampling for both datasets to achieve balance. Subsequent pre-processing generates datasets of different size and quality, all geared towards blood cell classification. We extend conventional evaluation tools with novel metrics—“accuracy difference” and “loss difference”—to detect overfitting or underfitting and evaluate their utility as potential indicators for learning behavior and promoting the self-confidence of ImageNet models. Our results show that these metrics effectively track learning progress and improve the reliability and overall performance of ImageNet models in new applications. This study highlights the transformative potential of turning ImageNet models into self-aware entities that significantly improve their robustness and efficiency in various AI tasks. This groundbreaking approach opens new perspectives for increasing the effectiveness of transfer learning in real-world AI implementations.

Efficient Storage and Analysis of Genomic Data: A k-mer Frequency Mapping and Image Representation Method.

k-mer frequencies are crucial for understanding DNA sequence patterns and structure, with applications in motif discovery, genome classification, and short read assembly. However, the exponential increase in the dimension of frequency tables with increasing k-mer length poses storage challenges. In this study, we present a novel method for compressing k-mer data without information loss, aiming to optimize storage and analysis processes. We employed Chaos Game Representation (CGR) to map k-mers to coordinates and used these components to generate raster images of k-mers. The CGR maps were partitioned and labeled based on substrings, with each substring mapped to a subframe, creating a fractal-like structure. The entire k-mer frequency set of each genomic sequence was represented as a single image, with each pixel corresponding to a specific k-mer and its occurrence. This approach reduced file size by up to 16-fold compared to plain text and 3-fold compared to binary format. Furthermore, we demonstrated the feasibility of performing alignment-free similarity analyses on images derived from k-mer frequencies of whole genome sequences from 14 plant species. Our results highlight the potential of this method as a fast and efficient tool for accessing, processing, and analyzing large biological sequence datasets, enabling the retrieval of k-mer frequencies and image reconstruction.

Retraction Note: Deep learning-based soft computing model for image classification application

Retraction Note: Deep learning-based soft computing model for image classification application

Deep Learning Applications in Natural Language Processing and Optimization Strategies

In recent years, deep learning has made significant advancements in the field of natural language processing (NLP), driving the development of language understanding and generation tasks. Deep learning utilizes multi-layer neural networks to automatically extract and represent complex data features. This paper first introduces the basic principles of deep learning and mainstream frameworks (such as TensorFlow and PyTorch) and discusses core NLP tasks, including word embedding, language modeling, and text generation. Subsequently, it analyzes specific practices of deep learning applications in text classification, machine translation, automatic summarization, dialogue systems, and speech recognition. Further discussion covers model optimization methods, including structural optimization (such as RNN, LSTM, GRU, and Transformer), data preprocessing and feature engineering, hyperparameter tuning, and accelerated computation. Finally, it explores cutting-edge optimization strategies like federated learning, model compression, self-supervised learning, and transfer learning, proposing future research directions and challenges. This paper aims to systematically analyze the applications and optimization strategies of deep learning in NLP.

Deep Learning Method of Precious Wood Image Classification Based on Microscopic Computed Tomography

Correctly identifying precious wood species is crucial for import and export trade and furniture material identification. This study utilizes nondestructive testing (Microscopic Computed Tomography, Micro-CT) to capture microscopic images of the transverse, radial, and tangential sections of 24 precious wood species, creating a comprehensive dataset. The SLConNet deep learning model is developed, enhancing recognition accuracy through multi-scale convolution and an improved residual block structure. The experiment results show that the classification accuracy of the transverse, radial and tangential sections is 98.72, 96.75 and 95.36 % respectively when the gain value is 0.8. The model outperforms traditional models like Alexnet, ResNet50, Inception-V3, and Xception. This research highlights the efficiency of nondestructive testing in obtaining a large number of microscopic wood images, compared to traditional anatomical methods. The SLConNet model showcases high accuracy in precision, recall, and specificity, suggesting its potential for widespread applications in wood classification.

Incorporation of Phase Change Materials in Buildings

This review paper explores the integration of phase change materials (PCMs) in building insulation systems to enhance energy efficiency and thermal comfort. Through an extensive analysis of existing literature, the thermal performance of PCM-enhanced building envelopes is evaluated under diverse environmental conditions. This review highlights that PCMs effectively moderate indoor temperatures by absorbing and releasing heat during phase transitions, maintaining a stable indoor climate. This paper also delves into the detailed concepts of PCMs, including their classification and various applications within building insulation. It is noted that different types of PCMs have unique thermal properties and potential uses, which can be tailored to specific building requirements and climatic conditions. Furthermore, cost–benefit and environmental assessments presented in the reviewed studies suggest that incorporating PCMs into building materials offers significant potential for reducing energy consumption and mitigating environmental impacts. These assessments indicate that PCMs can lead to substantial energy savings by decreasing the reliance on heating and cooling systems, thereby lowering overall energy costs and carbon emissions. However, despite the promising outlook, this review identifies a need for further research to optimize PCM formulations and integration methods. This optimization is essential for overcoming current challenges and facilitating the widespread adoption of PCMs in the construction industry. Addressing issues such as long-term durability, compatibility with existing building materials, and cost-effectiveness will be crucial for maximizing the benefits of PCMs in enhancing energy efficiency and sustainability in buildings. Overall, this review underscores the transformative potential of PCMs in building insulation practices. By providing a comprehensive overview of PCM classifications, applications, and their impacts on energy efficiency and environmental sustainability, this paper lays the groundwork for future advancements and research directions in the field of PCM-enhanced building technologies.

Geoinference of author affiliations using NLP-based text classification

Author affiliations are essential in bibliometric studies, requiring relevant information extraction from free-text affiliations. Precisely determining an author’s location from their affiliation is crucial for understanding research networks, collaborations, and geographic distribution. Existing geoparsing tools using regular expressions have limitations due to unstructured and ambiguous affiliations, resulting in erroneous location identification, especially for unconventional variations or misspellings. Moreover, their inefficient handling of big datasets hampers large-scale bibliometric studies. Though machine learning-based geoparsers exist, they depend on explicit location information, creating challenges when detailed geographic data is absent. To address these issues, we developed and evaluated a natural language processing model to predict the city, state, and country from an author’s free-text affiliation. Our model automates location inference, overcoming drawbacks of existing methods. Trained and tested with MapAffil, a publicly available geoparsed dataset of PubMed affiliations up to 2018, our model accurately retrieves high-resolution locations, even without explicit mentions of a city, state, or even country. Leveraging NLP techniques and the LinearSVC algorithm, our machine learning model achieves superior accuracy based on several validation datasets. This research demonstrates a practical application of text classification for inferring specific geographical locations from free-text affiliations, benefiting researchers and institutions in analyzing research output distribution.

A speech-based convolutional neural network for human body posture classification

Abstract Problem In recent years, computing and sensing advances have helped to develop efficient human posture classification systems, which assist in creating health systems that contribute to enhancing elder’s and disable’s life quality in context-aware and ambient assistive living applications. Other applications of body posture classification include the simulation of human bodies into virtual characters, security applications such as kidnapping and the body position of the kidnapper or victim, which can provide useful information to the negotiator or the rescue team, and other sports applications. Aim This work aims to propose a body posture classification system based on speech using deep learning techniques. Methods All samples pass through the preprocessing phase. The Mel-frequency cepstral coefficient (MFCC) with 12 coefficients was used as features, and the best features were selected by the correlation-based feature selection method. Two supervised learning techniques called artificial hydrocarbon networks (AHN) and convolutional neural networks (CNN) were deployed to efficiently classify body postures and movements. This research aims to detect six human positions through speech, which include walking downstairs, sitting, walking upstairs, running, laying, and walking. The dataset used in this work was prepared by the authors, named the human activity speech dataset. Results The best performance gained was with the AHN classifier of 68.9% accuracy and 67.1% accuracy with the CNN. Conclusion It was concluded that the MFCC features are the most powerful features in body position classification, and the deep learning methods are powerful in classifying body positions.

Application of ECG signal processing and classification based on CNN in wearable devices

Application of ECG signal processing and classification based on CNN in wearable devices

EEG Channel Estimation using CNN-Based Depression Classifier

Depression is a mental disorder that results to deteriorating effects in the lives of many people around the world. Traditional methods of diagnosing depression are based on interviews and questionaries. However, there were studies that have shown the possibilities of detecting depression biomarkers and perform classification. Electroencephalogram (EEG) analysis is one way of doing this. A series of signal processing techniques were used to streamline EEG data into a form which are recognizable by a machine learning algorithm. In this study, a convolutional neural network (CNN) – based depression classifier was used to classify depression using three EEG systems with 5, 16, and 128 channel locations. The aim is to estimate if there are differences in terms of classification performance. Results show that a system with locations for 5 and 16 EEG channels can achieve 98% accuracy like a 128-channel system. Hence, EEG systems with fewer number of electrodes can be utilized for depression classification applications.

A Memristor-Based Circuit with the Loser-Take-All Mechanism for Classification

Traditional multi-class classification circuits mostly use the mechanism of winner-take-all. In this paper, a memristor-based classification circuit with the loser-take-all mechanism is designed. The winner-take-all mechanism selects the most active neuron or signal while suppressing others, whereas the loser-take-all mechanism suppresses the most active and amplifies weaker signals. The goal of the loser-take-all mechanism is to determine which class an item does not belong to, rather than to determine which class the item belongs to. The loser-take-all mechanism can use relatively undemanding criteria to correctly classify the majority of categories that are misclassified by the winner-take-all mechanism. The designed circuit includes input modules, control modules and suppression modules which realize the multi-classification function based on the loser-take-all mechanism. The simulation results in Cadence show that the circuit can be used to realize complicated classification applications. The memristor-based classification circuit with the loser-take-all mechanism can capture the subtle nuances of various categories and provide a flexible approach to classification tasks.

An Improved Ant Colony Clustering Algorithm and Application in Classification of Tree Species

Abstract Clustering is one of the most challenging research topics and has attracted the attention of many researchers. Clustering aims to divide a given set of objects into several categories according to certain rules. It is widely applied to practical problems, such as knowledge discovery, artificial intelligence, scientific data analysis, business, biology, medical diagnosis, text mining, and web data mining and so on. Ant colony clustering algorithm (ACCA), which integrates ACA into clustering, is an efficient swarm intelligence algorithm. There are some weakness in the basic ant colony clustering algorithm, such as low solving efficiency, poor calculation accuracy and iteration stagnation. In this article, an improved Ant Colony Clustering Algorithm (IACCA), in which novel picking-up and dropping strategy is introduced, is proposed and applied to classification of tree species. The experiments indicate efficiency of IACCA for classification of tree species.

Revisiting typing systems for group B Streptococcus prophages: an application in prophage detection and classification in group B Streptococcus isolates from Argentina.

Group B Streptococcus (GBS) causes severe infections in neonates and adults with comorbidities. Prophages have been reported to contribute to GBS evolution and pathogenicity. However, no studies are available to date on the presence and diversity of prophages in GBS isolates from humans in South America. This study provides insights into the prophage content of 365 GBS isolates collected from clinical samples in the context of an Argentinean multicentric study. Using whole-genome sequence data, we implemented two previously proposed methods for prophage typing: a PCR approach (carried out in silico) coupled with a blastx-based method to classify prophages based on their prophage group and integrase type, respectively. We manually searched the genomes and identified 325 prophages. However, only 80% of prophages could be accurately categorized with the previous approaches. Integration of phylogenetic analysis, prophage group and integrase type allowed for all to be classified into 19 prophage types, which correlated with GBS clonal complex grouping. The revised prophage typing approach was additionally improved by using a blastn search after enriching the database with ten new genes for prophage group classification combined with the existing integrase typing method. This modified and integrated typing system was applied to the analysis of 615 GBS genomes (365 GBS from Argentina and 250 from public databases), which revealed 29 prophage types, including two novel integrase subtypes. Their characterization and comparative analysis revealed major differences in the lysogeny and replication modules. Genes related to bacterial fitness, virulence or adaptation to stressful environments were detected in all prophage types. Considering prophage prevalence, distribution and their association with bacterial virulence, it is important to study their role in GBS epidemiology. In this context, we propose the use of an improved and integrated prophage typing system suitable for rapid phage detection and classification with little computational processing.

Antipsychotic Drugs: A Concise Review of History, Classification, Indications, Mechanism, Efficacy, Side Effects, Dosing, and Clinical Application.

The introduction of the first antipsychotic drug, chlorpromazine, was a milestone for psychiatry. The authors review the history, classification, indications, mechanism, efficacy, side effects, dosing, drug initiation, switching, and other practical issues and questions related to antipsychotics. Classifications such as first-generation/typical versus second-generation/atypical antipsychotics are neither valid nor useful; these agents should be described according to the Neuroscience-based Nomenclature (NbN). Antipsychotic drugs are not specific for treating schizophrenia. They reduce psychosis regardless of the underlying diagnosis, and they go beyond nonspecific sedation. All currently available antipsychotic drugs are dopamine blockers or dopamine partial agonists. In schizophrenia, effect sizes for relapse prevention are larger than for acute treatment. A major unresolved problem is the implausible increase in placebo response in antipsychotic drug trials over the decades. Differences in side effects, which can be objectively measured, such as weight gain, are less equivocal than differences in rating-scale-measured (subjective) efficacy. The criteria for choosing among antipsychotics are mainly pragmatic and include factors such as available formulations, metabolism, half-life, efficacy, and side effects in previous illness episodes. Plasma levels help to detect nonadherence, and once-daily dosing at night (which is possible with many antipsychotics) and long-acting injectable formulations are useful when adherence is a problem. Dose-response curves for both acute treatment and relapse prevention follow a hyperbolic pattern, with maximally efficacious average dosages for schizophrenia of around 5 mg/day risperidone equivalents. Computer apps facilitating the choice between drugs are available. Future drug development should include pharmacogenetics and focus on drugs for specific aspects of psychosis.

An Expert Model Using Deep Learning for Image-based Pest Identification with the TSLM Approach for Enhancing Precision Farming

The rise of digital agriculture has sparked considerable interest in its potential to revolutionize farming practices by integrating of advanced technologies. Unmanned aerial vehicles (UAVs), commonly known as drones, have emerged as a crucial tool in precision agriculture, offering diverse applications for real-world scenarios. This research focuses on harnessing the power of drones in the context of digital agriculture support, particularly in addressing the challenges of crop protection and pest management. The objective is to develop an innovative approach for crop-disease diagnosis using an upgraded Transfer-Driven Self-Adaptive Learning Model (TSLM) that leverages multispectral remote sensing data of drone-captured images to improve pest classification and streamline pesticide application. The study explores nine potent deep neural network models' capabilities for identifying plant diseases using various approaches within the digital agriculture framework. Transfer learning and advanced feature extraction techniques are employed to tailor these deep neural networks to the specific crop protection context. Using of pre-trained deep learning models for feature extraction and fine-tuning enhances the effectiveness of the proposed model. Evaluating the model's performance, precision, sensitivity, specificity, and F1-score metrics are assessed, leading to the discovery that deep feature extraction and Self-Adaptive Learning Model (SLM) classification outperform traditional transfer learning methods. The novelty of this work lies in its application of communication protocols to coordinate a fleet of drones for crop protection within the digital agriculture framework. By combining deep learning techniques with transfer-driven self-adaptive learning, the proposed approach significantly enhances the accuracy and efficiency of pest classification in precision agriculture. The study's findings offer valuable insights into optimizing drone-based technologies to combat plant disease epidemics, thereby contributing to the advancement of digital agriculture and its role in supporting sustainable farming practices.

0.765-FJ/Bit/Search Content Addressable Memory Using Search Line Pre-Charge-Free (SLPF) Scheme

Content Addressable Memories (CAMs) are high-speed hardware lookup tables that are crucial in routing, packet classification, search, and other fast look-up applications. Despite having the advantage of high speed, they have limitations due to their power hungriness. This paper addresses the limitations of existing CAMs with two novel modifications in the proposed Search Line Pre-Charge-Free Low-Power-Content Addressable Memory (SLPFCAM) structure. The first novelty is their gated evaluation matching circuit and the second is a match line pre-charge controller. The proposed design reduces the energy consumption and improves the speed of the CAM by eliminating the SLprecharge phase in the CAM operation. The pre-charge controller prevents the attempt of match line pre-charge when it is not required. The design has been implemented using Cadence Virtuoso in a 90-nm CMOS technology with 1-V supply, and the post-layout simulation results show that the SLPFCAM makes significant improvements with a minimum of 83% less power consumption, a 91% less search delay, and the average energy consumption being 0.765 fJ/bit/search.