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

---

Enhancement of traffic forecasting through graph neural network-based information fusion techniques

Information Fusion, Год журнала: 2024, Номер 110, С. 102466 - 102466

Опубликована: Май 11, 2024

To improve forecasting accuracy and capture intricate interactions within transportation networks, information fusion approaches are crucial for traffic predictions based on graph neural networks (GNNs). GNNs offer a potentially effective framework capturing patterns among diverse elements, such as road segments crossings, by considering both temporal geographical dependencies. Although GNN-based has recently been investigated in many studies, there is need comprehensive reviews that examine predictions, including an analysis of their benefits challenges. This study addresses this knowledge gap offers future insights into the potential advancements developing fields research techniques, well implications urban planning smart cities. Existing demonstrates substantially enhanced techniques comparison to more conventional approaches. By integrating methods with GNNs, model capable spatial relationships between various locations network. Multi-source data integration models, social events, weather conditions, real-time sensor data, historical patterns. In addition, combining other AI like evolutionary algorithms or reinforcement learning could be efficient strategy. With combine best features several methods, hybrid models overall performance flexibility challenging situations.

Язык: Английский

---

Bi-directional information fusion-driven deep network for ship trajectory prediction in intelligent transportation systems

Transportation Research Part E Logistics and Transportation Review, Год журнала: 2024, Номер 192, С. 103770 - 103770

Опубликована: Сен. 20, 2024

Язык: Английский

---

Key technical indicators for stock market prediction

Machine Learning with Applications, Год журнала: 2025, Номер 20, С. 100631 - 100631

Опубликована: Март 1, 2025

Язык: Английский

---

Integrating BERT, GPT, Prophet Algorithm, and Finance Investment Strategies for Enhanced Predictive Modeling and Trend Analysis in Blockchain Technology

International Journal of Scientific Research in Computer Science Engineering and Information Technology, Год журнала: 2024, Номер 10(6), С. 1620 - 1645

Опубликована: Дек. 12, 2024

This paper explores the integration of advanced machine learning models, including BERT, GPT, and Prophet algorithm, with finance investment strategies to enhance predictive modeling trend analysis in blockchain technology. The rapid evolution has transformed financial ecosystems, offering decentralized platforms for secure transparent transactions. However, predicting market trends opportunities within this domain remains a complex challenge due high volatility multifaceted nature data. By leveraging natural language processing capabilities BERT GPT sentiment behavior prediction, combined time-series forecasting strength study aims provide robust framework analyzing blockchain-driven markets. Furthermore, ensures practical applicability by aligning insights real-world decision-making processes. proposed approach demonstrates potential optimizing portfolio management, enhancing risk mitigation, improving strategic ecosystems. work bridges gap between cutting-edge technologies innovation, valuable researchers practitioners both domains.

Язык: Английский

---

A Review of Time-Series Forecasting Algorithms for Industrial Manufacturing Systems

Machines, Год журнала: 2024, Номер 12(6), С. 380 - 380

Опубликована: Июнь 1, 2024

Time-series forecasting is crucial in the efficient operation and decision-making processes of various industrial systems. Accurately predicting future trends essential for optimizing resources, production scheduling, overall system performance. This comprehensive review examines time-series models their applications across diverse industries. We discuss fundamental principles, strengths, weaknesses traditional statistical methods such as Autoregressive Integrated Moving Average (ARIMA) Exponential Smoothing (ES), which are widely used due to simplicity interpretability. However, these often struggle with complex, non-linear, high-dimensional data commonly found To address challenges, we explore Machine Learning techniques, including Support Vector (SVM) Artificial Neural Network (ANN). These offer more flexibility adaptability, outperforming methods. Furthermore, investigate potential hybrid models, combine strengths different achieve improved prediction result accurate robust forecasts. Finally, newly developed generative Generative Adversarial (GAN) forecasting. emphasizes importance carefully selecting appropriate model based on specific industry requirements, characteristics, objectives.

Язык: Английский

---

A solar irradiance forecasting model using iterative filtering and bidirectional long short-term memory

Energy Sources Part A Recovery Utilization and Environmental Effects, Год журнала: 2024, Номер 46(1), С. 8202 - 8222

Опубликована: Июнь 26, 2024

Язык: Английский

---

Analysis of Conventional Feature Learning Algorithms and Advanced Deep Learning Models

Journal of Robotics Spectrum, Год журнала: 2023, Номер unknown, С. 1 - 12

Опубликована: Янв. 5, 2023

Representation learning or feature refers to a collection of methods employed in machine learning, which allows systems autonomously determine representations needed for classifications detection from unprocessed data. algorithms are specifically crafted acquire knowledge conceptual features that define The field state representation is centered on specific type involves the acquisition low-dimensional learned undergo temporal evolution and subject influence an agent's actions. Over past few years, deep architecture have been widely demonstrated exceptional performance various tasks, including but not limited object detection, speech recognition, image classification. This article provides comprehensive overview techniques data learning. Our research focuses examination conventional advanced models. paper presents introduction history, along with list available resources such as online courses, tutorials, books. Additionally, tool-boxes also provided further exploration this field. In conclusion, remarks future prospects

Язык: Английский

---

MFB: A Generalized Multimodal Fusion Approach for Bitcoin Price Prediction Using Time-Lagged Sentiment and Indicator Features

Expert Systems with Applications, Год журнала: 2024, Номер 261, С. 125515 - 125515

Опубликована: Окт. 19, 2024

Язык: Английский

---

A Comprehensive Study on Transformer-Based Time Series Forecasting

Advances in computational intelligence and robotics book series, Год журнала: 2025, Номер unknown, С. 159 - 180

Опубликована: Янв. 17, 2025

Time series forecasting is crucial for various real-world applications, such as energy consumption, traffic flow estimation, and financial market analysis. This chapter explores the application of deep learning models, specifically transformer-based models long-term time forecasting. Despite success transformers in sequence modeling, their permutation-invariant nature can lead to loss temporal information, posing challenges accurate Especially, embedding position-wise vector or time-stamp key long Another noted headache standard model squared computation complexity. studies development research field timer forecasting, challenging pain point, popular data sets, state-of-the-art benchmarks. The discussion covers implications, limitations, future directions, offering insights applying these advanced techniques problems.

Язык: Английский

---