A Comprehensive Review of Deep Learning: Architectures, Recent Advances, and Applications

Information, Год журнала: 2024, Номер 15(12), С. 755 - 755

Опубликована: Ноя. 27, 2024

Deep learning (DL) has become a core component of modern artificial intelligence (AI), driving significant advancements across diverse fields by facilitating the analysis complex systems, from protein folding in biology to molecular discovery chemistry and particle interactions physics. However, field deep is constantly evolving, with recent innovations both architectures applications. Therefore, this paper provides comprehensive review DL advances, covering evolution applications foundational models like convolutional neural networks (CNNs) Recurrent Neural Networks (RNNs), as well such transformers, generative adversarial (GANs), capsule networks, graph (GNNs). Additionally, discusses novel training techniques, including self-supervised learning, federated reinforcement which further enhance capabilities models. By synthesizing developments identifying current challenges, insights into state art future directions research, offering valuable guidance for researchers industry experts.

Язык: Английский

---

A Hybrid Deep Learning Approach with Generative Adversarial Network for Credit Card Fraud Detection

Technologies, Год журнала: 2024, Номер 12(10), С. 186 - 186

Опубликована: Окт. 2, 2024

Credit card fraud detection is a critical challenge in the financial industry, with substantial economic implications. Conventional machine learning (ML) techniques often fail to adapt evolving patterns and underperform imbalanced datasets. This study proposes hybrid deep framework that integrates Generative Adversarial Networks (GANs) Recurrent Neural (RNNs) enhance capabilities. The GAN component generates realistic synthetic fraudulent transactions, addressing data imbalance enhancing training set. discriminator, implemented using various DL architectures, including Simple RNN, Long Short-Term Memory (LSTM) networks, Gated Units (GRUs), trained distinguish between real transactions further fine-tuned classify as or legitimate. Experimental results demonstrate significant improvements over traditional methods, GAN-GRU model achieving sensitivity of 0.992 specificity 1.000 on European credit dataset. work highlights potential GANs combined architectures provide more effective adaptable solution for detection.

Язык: Английский

---

Applications of Long Short-Term Memory (LSTM) Networks in Polymeric Sciences: A Review

Polymers, Год журнала: 2024, Номер 16(18), С. 2607 - 2607

Опубликована: Сен. 14, 2024

This review explores the application of Long Short-Term Memory (LSTM) networks, a specialized type recurrent neural network (RNN), in field polymeric sciences. LSTM networks have shown notable effectiveness modeling sequential data and predicting time-series outcomes, which are essential for understanding complex molecular structures dynamic processes polymers. delves into use models polymer properties, monitoring polymerization processes, evaluating degradation mechanical performance Additionally, it addresses challenges related to availability interpretability. Through various case studies comparative analyses, demonstrates different science applications. Future directions also discussed, with an emphasis on real-time applications need interdisciplinary collaboration. The goal this is connect advanced machine learning (ML) techniques science, thereby promoting innovation improving predictive capabilities field.

Язык: Английский

---

Deep Learning in Finance: A Survey of Applications and Techniques

AI, Год журнала: 2024, Номер 5(4), С. 2066 - 2091

Опубликована: Окт. 28, 2024

Machine learning (ML) has transformed the financial industry by enabling advanced applications such as credit scoring, fraud detection, and market forecasting. At core of this transformation is deep (DL), a subset ML that robust in processing analyzing complex large datasets. This paper provides comprehensive overview key models, including Convolutional Neural Networks (CNNs), Long Short-Term Memory networks (LSTMs), Deep Belief (DBNs), Transformers, Generative Adversarial (GANs), Reinforcement Learning (Deep RL). Beyond summarizing their mathematical foundations processes, study offers new insights into how these models are applied real-world contexts, highlighting specific advantages limitations tasks algorithmic trading, risk management, portfolio optimization. It also examines recent advances emerging trends alongside critical challenges data quality, model interpretability, computational complexity. These can guide future research directions toward developing more efficient, robust, explainable address evolving needs sector.

Язык: Английский

---

Application of deep learning for high-throughput phenotyping of seed: a review

Artificial Intelligence Review, Год журнала: 2025, Номер 58(3)

Опубликована: Янв. 6, 2025

Abstract Seed quality is of great importance for agricultural cultivation. High-throughput phenotyping techniques can collect magnificent seed information in a rapid and non-destructive manner. Emerging deep learning technology brings new opportunities effectively processing massive diverse data from seeds evaluating their quality. This article comprehensively reviews the principle several high-throughput non-destructively collection information. In addition, recent research studies on application learning-based approaches inspection are reviewed summarized, including variety classification grading, damage detection, components prediction, cleanliness, vitality assessment, etc. review illustrates that combination be promising tool various phenotype seeds, which used effective evaluation industrial practical applications, such as breeding, management, selection food source.

Язык: Английский

---

Lunar Calendar Usage to Improve Forecasting Accuracy Rainfall via Machine Learning Methods

Applied Sciences, Год журнала: 2025, Номер 15(2), С. 675 - 675

Опубликована: Янв. 11, 2025

The lunar calendar is often overlooked in time-series data modeling despite its importance understanding seasonal patterns, as well economics, natural phenomena, and consumer behavior. This study aimed to investigate the effectiveness of forecasting rainfall levels using various machine learning methods. methods employed included long short-term memory (LSTM) gated recurrent unit (GRU) models test accuracy forecasts based on compared those Gregorian calendar. results indicated that incorporating generally provided greater for periods 3, 4, 6, 12 months model demonstrated higher prediction, exhibiting smaller errors (MAPE MBE values), whereas yielded somewhat larger tended underestimate values. These findings contributed advancement techniques, learning, adaptation non-Gregorian systems while also opening new opportunities further research into applications across domains.

Язык: Английский

---

Leveraging Advanced NLP Techniques and Data Augmentation to Enhance Online Misogyny Detection

Applied Sciences, Год журнала: 2025, Номер 15(2), С. 856 - 856

Опубликована: Янв. 16, 2025

Online misogyny is a significant societal challenge that reinforces gender inequalities and discourages women from engaging fully in digital spaces. Traditional moderation methods often fail to address the dynamic context-dependent nature of misogynistic language, making adaptive solutions essential. This study presents framework integrates advanced natural-language processing techniques with strategic data augmentation improve detection content. Key contributions include emoji decoding interpret symbolic communication, contextual expansion using Sentence-Transformer models, LDA-based topic modeling enhance richness understanding. The incorporates machine-learning, deep-learning, Transformer-based models handle complex nuanced language. Performance analysis highlights effectiveness selected comparative results emphasize transformative role augmentation. significantly enhanced model robustness, improved generalization, strengthened

Язык: Английский

---

Progress and Limitations in Forest Carbon Stock Estimation Using Remote Sensing Technologies: A Comprehensive Review

Forests, Год журнала: 2025, Номер 16(3), С. 449 - 449

Опубликована: Март 2, 2025

Forests play a key role in carbon sequestration and oxygen production. They significantly contribute to peaking neutrality goals. Accurate estimation of forest stocks is essential for precise understanding the capacity ecosystems. Remote sensing technology, with its wide observational coverage, strong timeliness, low cost, stock research. However, challenges data acquisition processing include variability, signal saturation dense forests, environmental limitations. These factors hinder accurate estimation. This review summarizes current state research on from two aspects, namely remote methods, highlighting both advantages limitations various sources models. It also explores technological innovations cutting-edge field, focusing deep learning techniques, optical vegetation thickness impact forest–climate interactions Finally, discusses including issues related quality, model adaptability, stand complexity, uncertainties process. Based these challenges, paper looks ahead future trends, proposing potential breakthroughs pathways. The aim this study provide theoretical support methodological guidance researchers fields.

Язык: Английский

---

Evaluating Machine Learning-Based Soft Sensors for Effluent Quality Prediction in Wastewater Treatment Under Variable Weather Conditions

Sensors, Год журнала: 2025, Номер 25(6), С. 1692 - 1692

Опубликована: Март 8, 2025

Maintaining effluent quality in wastewater treatment plants (WWTPs) comes with significant challenges under variable weather conditions, where sudden changes flow rate and increased pollutant loads can affect performance. Traditional physical sensors became both expensive susceptible to failure extreme conditions. In this study, we evaluate the performance of soft based on artificial intelligence (AI) predict components underlying calculation index (EQI). We thus focus our study three ML models: Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) Transformer. Using Benchmark Simulation Model no. 2 (BSM2) as WWTP, were able obtain datasets for training models their dry scenarios, rainy episodes, storm events. To improve classification networks according type weather, developed a Random Forest (RF)-based meta-classifier. The results indicate that conditions Transformer network achieved best performance, while rain episodes scenarios GRU was capture variations highest accuracy. LSTM performed normally stable but struggled rapid fluctuations. These support decision integrate AI-based predictive WWTPs, highlighting top performances recurrent feed-forward (Transformer) obtaining predictions different

Язык: Английский

---

Air Traffic Trends and UAV Safety: Leveraging Automatic Dependent Surveillance–Broadcast Data for Predictive Risk Mitigation

Aerospace, Год журнала: 2025, Номер 12(4), С. 284 - 284

Опубликована: Март 28, 2025

With the significant potential of Unmanned Aircraft Vehicles (UAVs) extending throughout various fields and industries, their proliferation raises concerns regarding risks within national airspace system (NAS). To enhance safe efficient integration UAVs into airport environments, this paper presents an analysis temporal statistical patterns in flight traffic, predictive modeling future traffic trends using machine learning, identification optimal time windows for UAV operations airports. The framework was developed historical Automatic Dependent Surveillance–Broadcast (ADS-B) data obtained from OpenSky Network. Historical Class B, C, D airports California are processed, is carried out to identify variations including daily, weekly, seasonal trends. A recurrent neural network (RNN) model incorporating Long Short-Term Memory (LSTM) architecture forecast counts based on patterns, achieving mean absolute error (MAE) values 4.52, 2.13, 0.87 airports, respectively. findings highlight distinct across classes, emphasizing practicality utilizing ADS-B scheduling minimize conflicts with manned aircraft. Additionally, study explores influence external factors, weather conditions dataset limitations prediction accuracy. By integrating learning real-time data, research provides a optimizing operations, supporting management improving regulatory compliance controlled airspace.

Язык: Английский

---