A systematic review of hyperparameter optimization techniques in Convolutional Neural Networks

Decision Analytics Journal, Год журнала: 2024, Номер 11, С. 100470 - 100470

Опубликована: Апрель 24, 2024

Convolutional Neural Network (CNN) is a prevalent topic in deep learning (DL) research for their architectural advantages. CNN relies heavily on hyperparameter configurations, and manually tuning these hyperparameters can be time-consuming researchers, therefore we need efficient optimization techniques. In this systematic review, explore range of well used algorithms, including metaheuristic, statistical, sequential, numerical approaches, to fine-tune hyperparameters. Our offers an exhaustive categorization (HPO) algorithms investigates the fundamental concepts CNN, explaining role variants. Furthermore, literature review HPO employing above mentioned undertaken. A comparative analysis conducted based strategies, error evaluation accuracy results across various datasets assess efficacy methods. addition addressing current challenges HPO, our illuminates unresolved issues field. By providing insightful evaluations merits demerits objective assist researchers determining suitable method particular problem dataset. highlighting future directions synthesizing diversified knowledge, survey contributes significantly ongoing development optimization.

Язык: Английский

---

Next–Generation Intrusion Detection for IoT EVCS: Integrating CNN, LSTM, and GRU Models

Mathematics, Год журнала: 2024, Номер 12(4), С. 571 - 571

Опубликована: Фев. 14, 2024

In the evolving landscape of Internet Things (IoT) and Industrial IoT (IIoT) security, novel efficient intrusion detection systems (IDSs) are paramount. this article, we present a groundbreaking approach to for IoT-based electric vehicle charging stations (EVCS), integrating robust capabilities convolutional neural network (CNN), long short-term memory (LSTM), gated recurrent unit (GRU) models. The proposed framework leverages comprehensive real-world cybersecurity dataset, specifically tailored IIoT applications, address intricate challenges faced by EVCS. We conducted extensive testing in both binary multiclass scenarios. results remarkable, demonstrating perfect 100% accuracy classification, an impressive 97.44% six-class 96.90% fifteen-class setting new benchmarks field. These achievements underscore efficacy CNN-LSTM-GRU ensemble architecture creating resilient adaptive IDS infrastructures. algorithm, accessible via GitHub, represents significant stride fortifying EVCS against diverse array threats.

Язык: Английский

---

Advanced Temporal Deep Learning Framework for Enhanced Predictive Modeling in Industrial Treatment Systems

Results in Engineering, Год журнала: 2025, Номер unknown, С. 104158 - 104158

Опубликована: Янв. 1, 2025

Язык: Английский

---

Online Machine Learning for Intrusion Detection in Electric Vehicle Charging Systems

Mathematics, Год журнала: 2025, Номер 13(5), С. 712 - 712

Опубликована: Фев. 22, 2025

Electric vehicle (EV) charging systems are now integral to smart grids, increasing the need for robust and scalable cyberattack detection. This study presents an online intrusion detection system that leverages Adaptive Random Forest classifier with Windowing drift identify real-time evolving threats in EV infrastructures. The is evaluated using real-world network traffic from CICEVSE2024 dataset, ensuring practical applicability. For binary detection, model achieves 0.9913 accuracy, 0.9999 precision, 0.9914 recall, F1-score of 0.9956, demonstrating highly accurate threat It effectively manages concept drift, maintaining average accuracy 0.99 during events. In multiclass attains 0.9840 0.9831 event 0.96. computationally efficient, processing each instance just 0.0037 s, making it well-suited deployment. These results confirm machine learning methods can secure source code publicly available on GitHub, reproducibility fostering further research. provides a efficient cybersecurity solution protecting networks threats.

Язык: Английский

---

A Novel Approach for Real-Time Server-Based Attack Detection Using Meta-Learning

IEEE Access, Год журнала: 2024, Номер 12, С. 39614 - 39627

Опубликована: Янв. 1, 2024

Modern networks are crucial for seamless connectivity but face various threats, including disruptive network attacks, which can result in significant financial and reputational risks. To counter these challenges, AI-based techniques being explored protection, requiring high-quality datasets training. In this study, we present a novel methodology utilizing Ubuntu Base Server to simulate virtual environment real-time collection of attack datasets. By employing Kali Linux as the attacker machine Wireshark data capture, compile Server-based Network Attack (SNA) dataset, showcasing UDP, SYN, HTTP flood attacks. Our primary goal is provide publicly accessible, server-focused dataset tailored research. Additionally, leverage advanced AI methods detection proposed meta-RF-GNB (MRG) model combines Gaussian Naive Bayes Random Forest predictions, achieving an impressive accuracy score 99.99%. We validate efficiency MRG using cross-validation, obtaining notable mean 99.94% with minimal standard deviation 0.00002. Furthermore, conducted statistical t-test evaluate significance compared other top-performing models.

Язык: Английский

---

Intrusion detection systems for IoT based on bio-inspired and machine learning techniques: a systematic review of the literature

Cluster Computing, Год журнала: 2024, Номер 27(7), С. 8655 - 8681

Опубликована: Апрель 14, 2024

Язык: Английский

---

Explainable TabNet Transformer-based on Google Vizier Optimizer for Anomaly Intrusion Detection System

Knowledge-Based Systems, Год журнала: 2025, Номер unknown, С. 113351 - 113351

Опубликована: Март 1, 2025

Язык: Английский

---

Artificial intelligence-based non-invasive bilirubin prediction for neonatal jaundice using 1D convolutional neural network

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Апрель 4, 2025

Язык: Английский

---

Assessing landscape ecological vulnerability to riverbank erosion in the Middle Brahmaputra floodplains of Assam, India using machine learning algorithms

CATENA, Год журнала: 2023, Номер 234, С. 107581 - 107581

Опубликована: Окт. 9, 2023

Язык: Английский

---

Enhancing Network Intrusion Detection Through the Application of the Dung Beetle Optimized Fusion Model

IEEE Access, Год журнала: 2024, Номер 12, С. 9483 - 9496

Опубликована: Янв. 1, 2024

With the rapid development of information communication and mobile device technologies, smart devices have become increasingly popular, providing convenience to households enhancing level intelligence in daily life. This trend is also driving innovation progress various fields, including healthcare, transportation, industry. However, as technology continues proliferate, network security concerns prominent, making protection digital life data an urgent priority. Intrusion detection has always played important role field security. Traditional intrusion systems predominantly rely on anomaly identify potential intrusions by detecting abnormal patterns traffic. technological advancements, machine learning-based methods emerged cornerstone modern detection, enabling more precise identification behaviors learning normal In response these challenges, this paper introduces innovative model that amalgamates Attention-CNN-BiLSTM (ACBL) Temporal Convolutional Network (TCN) architectures. The ACBL TCN models excel processing spatial temporal features within traffic data, respectively. integration harnesses diverse neural structures elevate overall performance accuracy. Furthermore, a unique approach inspired dung beetles' natural behavior, incorporating Tent mapping-enhanced Dung Beetle Optimization Algorithm (TDBO), leveraged for both optimizing feature selection parameters searching optimal hyperparameters. obtained from TDBO are then combined with importance ranking Random Forest algorithm, ensuring can be better selected enhance performance. novel model, TDBO-ACBLT validates its using UNSW-NW15 dataset. excels compared common algorithms achieves superior parameter optimization accuracy over Harris's Hawk (HHO), Particle Swarm (PSO), (DBO). proposed higher than prevalent models.

Язык: Английский

---