Deep Learning Attention Mechanism in Medical Image Analysis: Basics and Beyonds

Li, Xiang; Li, Minglei; Yan, Pingkun; Li, Guanyi; Jiang, Yuchen; Luo, Hao; Yin, Shen

doi:10.53941/ijndi0201006

Cited by 68 publications

(22 citation statements)

References 117 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One is that the comparison test data set used is only 200million parameters, which may not fully reflect the specific performance of each model. In the following research, more languages will be used for testing on larger datasets to fully reflect the specific performance of each model, and more optimization methods used in other studies [12,13] will be explored to maximize the performance of the model.…”

Section: Discussionmentioning

confidence: 99%

Investigation on task effect analysis and optimization strategy of multimodal large model based on Transformers architecture for various languages

Zhang

2024

ACE

View full text Add to dashboard Cite

As artificial intelligence technology advances swiftly, the Transformers architecture has emerged as a pivotal model for handling multimodal data. This investigation delves into the impact of multimodal large-scale models utilizing the Transformers architecture for addressing various linguistic tasks, along with proposing optimization approaches tailored to this context. Through a series of experiments, this study scrutinized the performance of these models on multilingual datasets, engaging in a comprehensive analysis of the key determinants influencing their effectiveness. Firstly, several models of transformers architecture are pre trained on the same corpus, including ERNIE, GPT, ViT, VisualBERT, and a series of tests are carried out on these models in English, Chinese, Spanish and other languages. By comparing the performance of different models, it is found that these models show significant performance differences when dealing with tasks in different languages. Further, through analysis and experimental verification, this paper proposes a series of optimization strategies for different languages, including: annotation method for language specific datasets, incremental fine-tuning method for tuning, increasing the size of datasets, using multi task learning, etc. Experiments show that these methods have achieved remarkable results, and put forward the future research direction.

show abstract

Section: Discussionmentioning

confidence: 99%

Investigation on task effect analysis and optimization strategy of multimodal large model based on Transformers architecture for various languages

Zhang

2024

ACE

View full text Add to dashboard Cite

show abstract

“…RNNs were utilized by Zhou et al ( 2023 ) to model texts by taking into account their temporal information. Li et al ( 2023 ) achieved outstanding results in a sentiment classification test by modeling utterances using a tree LSTM model to approximate the sentence structure. By segmenting a text according to sentences, obtaining vectors through convolutional pooling operation, and then inputting them into LSTM according to temporal relations to construct a CNN-LSTM model and apply it to the task of sentiment analysis, Alirezazadeh et al ( 2023 ) primarily addressed the issue of temporal and long-range dependencies in a chapter-level text.…”

Section: Related Studiesmentioning

confidence: 99%

Prediction of emotion distribution of images based on weighted K-nearest neighbor-attention mechanism

Cheng

2024

Front. Comput. Neurosci.

View full text Add to dashboard Cite

Existing methods for classifying image emotions often overlook the subjective impact emotions evoke in observers, focusing primarily on emotion categories. However, this approach falls short in meeting practical needs as it neglects the nuanced emotional responses captured within an image. This study proposes a novel approach employing the weighted closest neighbor algorithm to predict the discrete distribution of emotion in abstract paintings. Initially, emotional features are extracted from the images and assigned varying K-values. Subsequently, an encoder-decoder architecture is utilized to derive sentiment features from abstract paintings, augmented by a pre-trained model to enhance classification model generalization and convergence speed. By incorporating a blank attention mechanism into the decoder and integrating it with the encoder's output sequence, the semantics of abstract painting images are learned, facilitating precise and sensible emotional understanding. Experimental results demonstrate that the classification algorithm, utilizing the attention mechanism, achieves a higher accuracy of 80.7% compared to current methods. This innovative approach successfully addresses the intricate challenge of discerning emotions in abstract paintings, underscoring the significance of considering subjective emotional responses in image classification. The integration of advanced techniques such as weighted closest neighbor algorithm and attention mechanisms holds promise for enhancing the comprehension and classification of emotional content in visual art.

show abstract

“…The creation of hybrid models, which combine the advantages of many architectural components to produce superior outcomes, has been one of the most significant movements in recent years. In this research [7], the Efficient Channel Attention Network (ECA-Net), ResNet50, and DenseNet201 are three well-known architectures that are combined in a novel hybrid model is presented. This fusion seeks to achieve hitherto unseen levels of efficacy and accuracy in image recognition tasks by utilizing the complementing qualities of two architectures.…”

Section: Introductionmentioning

confidence: 99%

Palm Leaf Health Management: A Hybrid Approach for Automated Disease Detection and Therapy Enhancement

Nobel,

Imran,

Bina

et al. 2024

IEEE Access

View full text Add to dashboard Cite

Deep Learning and computer vision have become potent agricultural technologies in recent years. These technologies are essential for identifying hazardous plant leaf diseases, which significantly impact crop quality and productivity. The precise distinction between healthy and damaged palm leaves is at the core of this research. Our study marks a significant improvement in the area by introducing a novel method for identifying palm leaf disease using a hybrid model. Our hybrid model's central component combines the Efficient Channel Attention Network (ECA-Net) with reliable transfer learning techniques utilizing ResNet50 and DenseNet201. In addition to improving disease diagnosis accuracy, this fusion sets a new performance bar compared to earlier models. Our hybrid model maintains a validation accuracy of 98.67% while achieving an amazing 99.54% training accuracy in precisely identifying diseases. Compared to its contemporaries, it also performs exceptionally well in F1 score values, highlighting its remarkable prowess in agricultural technology. This research provides a breakthrough method for disease detection in palm leaves. It will revolutionize the agriculture sector. INDEX TERMS Palm leaf disease, deep learning, CNN, transfer learning, hybrid model, k-fold, disease detection, automated diagnosis, agricultural sustainability.

show abstract

Deep Learning Attention Mechanism in Medical Image Analysis: Basics and Beyonds

Cited by 68 publications

References 117 publications

Investigation on task effect analysis and optimization strategy of multimodal large model based on Transformers architecture for various languages

Investigation on task effect analysis and optimization strategy of multimodal large model based on Transformers architecture for various languages

Prediction of emotion distribution of images based on weighted K-nearest neighbor-attention mechanism

Palm Leaf Health Management: A Hybrid Approach for Automated Disease Detection and Therapy Enhancement

Contact Info

Product

Resources

About