UzADL: Anomaly detection and localization using graph Laplacian matrix-based unsupervised learning method

Olimov, Bekhzod; Veluvolu, Kalyana C.; Paul, Anand; Kim, Jeonghong

doi:10.1016/j.cie.2022.108313

Cited by 17 publications

(6 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table V compares the performance of our proposed bacteria classification models with other popular deep learning models including AlexNet [21], VGG [26], ResNet [22], DenseNet [27], SqueezeNet [28], vision transformer (ViT) [29], model soup [30] and Lion Fine-tune CNN [31]. Furthermore, we also compared the proposed method with the other two recent SOAT methods namely UzADL [32] and CMSFL [33] which used WIB-ReLU. As mentioned earlier, even though each model has its own based architecture, all of them were developed using the same parameter setting to maintain a fair basis for comparison.…”

Section: Resultsmentioning

confidence: 99%

“…Besides the aforementioned models, we also conduct a comparison of our proposed model with the most recent state-of-the-art (SOTA) image classification methods including vision transformer (ViT) [29], model soup [30], Lion Fine-tune CNN [31], UzADL [32] and CMSFL [33]. For such models, we regenerate the code and train using our bacteria dataset.…”

Section: ) Schedule Lengthmentioning

confidence: 99%

See 1 more Smart Citation

A Deep Learning Model for Bacterial Classification Using Big Transfer (BiT)

Visitsattaponge,

Bunkum,

Pintavirooj

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Identification and classification of bacterial genera and species are very important for medical prevention, diagnosis, and treatment. However, due to microbial diversity and high variability in appearance, the manual classification of bacteria is a challenging and time-consuming task. This paper aims to facilitate such a troublesome task using deep learning techniques. Through the utilization of a deep learning model, specifically a Big Transfer (BiT) combined with Graph Laplacian-based data cleaning and weight initialization based-rectified linear unit (WIB-Relu) activation, we have developed an accurate bacteria classification model. We have tested our proposed method on a public dataset of microscopic bacteria images, called the Digital Images of Bacteria Species (DIBaS), and achieved promising results with an accuracy of 99.11%, precision of 99.31%, recall of 99.09%, and F1 score of 99.06%, respectively. Moreover, the proposed bacteria classification performed well regardless of the size of the training data. We investigated its generalizability not only on the original dataset but also on the few shots (5-shots, 2-shots, and 1-shot) and augmented datasets.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: ) Schedule Lengthmentioning

confidence: 99%

A Deep Learning Model for Bacterial Classification Using Big Transfer (BiT)

Visitsattaponge,

Bunkum,

Pintavirooj

et al. 2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Image classification is a fundamental application within deep learning and computer vision, focusing on the task of categorizing input images into predetermined classes by extracting and encoding relevant image features for subsequent processing. In contrast to image segmentation and detection, which employ different techniques image classification holds significant importance within the realm of deep learning, with medical image classification being of particular concern [2,3]. Even a minor error in this domain could have severe consequences for human lives, underscoring the critical role of ongoing enhancements in advancing this field.…”

Section: Related Workmentioning

confidence: 99%

Detailed Comparative Analysis of Transfer Learning Based Ensemble Models for Mushroom Classification

Karegowda,

Bhat

et al. 2024

Preprint

View full text Add to dashboard Cite

Identifying mushroom species accurately is often challenging due to the vast diversity and visual differences among species at different stages of growth. This study investigates the effectiveness of stacking ensemble technique using a combination of 3–4 transfer learning models as Base classifiers with simple average and weighted average method, to enhance mushroom classification accuracy. Our research focuses on two primary aspects: the performance of individual transfer learning models and the impact of stacking ensemble techniques on transfer learning models for mushroom classification. The results of each ensemble of transfer learning model exhibited strong performance, achieving high accuracy in the range of 90.54% − 98.42% for Weighted Average models, with the best performance using three transfer learning modes: ResNet50V2-MobileNet–VGG16. The weighted average approach assigned weights to each model based on their class wise performance during model construction and henceforth results in better accuracy compared to simple average method.

show abstract

“…This method enables effective differentiation between normal traffic and malicious behavior . Olimov et al [55] proposed a novel unsupervised learning method based on graph Laplacian matrix for anomaly detection and localization. This method effectively detects and localizes anomalies using graph structural information without relying on labeled anomaly data.…”

Section: Introductionmentioning

confidence: 99%

BBO-CFAT: Network Intrusion Detection Model Based on BBO Algorithm and Hierarchical Transformer

Jiang,

Fu,

Wang

2024

IEEE Access

View full text Add to dashboard Cite

In today's network environments vulnerable to cyber threats like hackers and viruses, intrusion detection technology is considered the most effective means of detection and defense. Deep neural networks are commonly used in intrusion detection technology. However, improving the model's ability to extract feature information and reducing computational space while retaining local feature information are critical challenges that need to be addressed.To tackle these issues, this paper proposes a model named BBO-CFAT, which combines the Biogeography-Based Optimization algorithm (BBO) for feature selection and an improved Transformer model for preserving context information and reducing computational space. Specifically, the BBO-CFAT model employs a roulette selection method to control the operations of migration and mutation operators. It utilizes feature information entropy to weight updates of adaptive variables in these operators, thereby enhancing the credibility of feature selection. Furthermore, the Transformer framework is hierarchically designed to facilitate the acquisition of context information. Additionally, depthwise separable convolutions are employed to reduce computational space, thereby improving computational efficiency and training speed.Experimental evaluations using the CIC-IDS2017 and NSL-KDD datasets demonstrate promising accuracies for BBO-CFAT on both datasets, achieving 99.1% and 97.5% accuracy, respectively, surpassing the performance of comparative experiments.Overall, the BBO-CFAT model provides a comprehensive solution to the challenges of intrusion detection, effectively balancing feature preservation, computational efficiency, and training accuracy.

show abstract

UzADL: Anomaly detection and localization using graph Laplacian matrix-based unsupervised learning method

Cited by 17 publications

References 33 publications

A Deep Learning Model for Bacterial Classification Using Big Transfer (BiT)

A Deep Learning Model for Bacterial Classification Using Big Transfer (BiT)

Detailed Comparative Analysis of Transfer Learning Based Ensemble Models for Mushroom Classification

BBO-CFAT: Network Intrusion Detection Model Based on BBO Algorithm and Hierarchical Transformer

Contact Info

Product

Resources

About