Intelligent Systems with Applications最新文献

Medical research plays a crucial role within scientific research. Technological advancements, especially those related to the rise of machine learning, pave the way for the exploration of medical issues that were once beyond reach. Unstructured textual data, such as correspondence between doctors, operative reports, etc., often serve as a starting point for many medical applications.

However, for obvious privacy reasons, researchers do not legally have the right to access these documents as long as they contain sensitive data, as defined by regulations like GDPR (General Data Protection Regulation) or HIPAA (Health Insurance Portability and Accountability Act). De-identification, meaning the detection, removal or substitution of all sensitive information, is therefore a necessary step to facilitate the sharing of these data between the medical field and research. Over the past decade, various approaches have been proposed to de-identify medical textual data. However, while entity detection is a well-known task in the natural language processing field, it presents some specific challenges in the medical context. Moreover, existing substitution methods proposed in the literature often pay little attention to the medical relevance of de-identified data or are not very resilient to attacks.

This paper addresses these challenges. Firstly, an efficient system for detecting sensitive entities in French medical data and then accurately substitute them was implemented. Secondly, robust strategies for generating substitutes that incorporate the medical utility of the data were provided, thereby minimizing the difference in utility between the original and de-identified data, and that mathematically ensure privacy protection. Thirdly, the utility of the de-identification system in a context of ICD-10 code association was evaluated. Finally, various systems developed to tackle ICD-10 code association were presented while providing a state-of-the-art model in French.

Generalisation across multiple tasks is a major challenge in deep learning for medical imaging applications, as it can cause a catastrophic forgetting problem. One commonly adopted approach to address these challenges is to train the model from scratch, incorporating old and new data, classes, and tasks. However, this solution comes with its downsides, as it is time-consuming, requires high computational resources, is susceptible to bias, and lacks flexibility. To effectively address these issues, this paper introduces a generalisable DL framework that consists of three key components: self-supervised learning, feature fusion of a single task, and feature fusion of new classes or tasks. Using the proposed framework, DL models with the SVM classifier can accurately detect abnormalities in X-ray tasks, including the humerus and wrist, achieving an accuracy of 92.71% and 90.74%, respectively. These results were achieved using a single classifier with minimal training requirements when new tasks were introduced. Another experiment was performed on chest X-rays, where new classes were added to the pre-existing ones. Without requiring retraining with both old and new classes, our framework achieved a combined class accuracy of 98.18%. This demonstrates that the model has not forgotten the old data. The proposed framework enhances performance and brings flexibility and efficiency to the training process, saving time and computational resources.

Latent palmprints are integral to crime scene investigations, constituting a significant portion of encountered prints. They often suffer from poor ridge impressions, noise, and pronounced creases, setting them apart from other palmprint types. While progressive enhancement techniques are widely used for fingerprints, palmprints with numerous thick creases and larger sizes benefit more from region-growing techniques. Frequency domain-based palmprint enhancement excels in separating creases from ridges and reshaping ridge structures accurately. The key challenge lies in identifying suitable initial blocks for both region-growing and iterative enhancement techniques. Existing frequency domain-based quality maps, primarily designed for fingerprints, exhibit limited performance when applied to palmprints, especially latent ones. To address these issues, this paper introduces a new approach that combines region-growing and frequency domain-based enhancement techniques to improve latent palmprints. Our method leverages high-quality blocks, employs the orientation field obtained in the frequency domain to correct possible orientation errors in starting blocks, and utilizes varying weights to enhance all block types effectively. The experimental results indicate that the proposed approach surpasses the existing state-of-the-art techniques in terms of recognition accuracy.

Due to factors such as changing customer demands and supply disruptions, product design changes are inevitable during the product development process. Selecting an appropriate product change propagation path not only maintains product performance but also reduces generation time and cost. This paper investigates an intelligent optimization method for complex product change propagation paths. Firstly, considering change duration, change cost, and the impact degree on product performance, a parameter network model of the product is constructed based on the parameter linkage relationship between product parts. Secondly, to solve this model a change propagation path optimization method based on an improved ant colony algorithm is proposed. Finally, the effectiveness of the proposed model and algorithm is validated on the design change problem of TV products at Skyworth RGB Co., Ltd. Experimental results demonstrate that the proposed algorithm can generate highly competitive optimal change paths for TV products.

As people's environmental awareness improves, the development of green economy has become an important goal for countries to achieve sustainable development. The extensive generation of greenhouse gases during the cold chain logistics and distribution process poses a significant challenge to the advancement of the green economy. By utilizing an enhanced genetic algorithm (GA), this research examines the expenses associated with distributing fresh agricultural products via the cold chain and constructs a model for optimizing the path of distribution. Simulation experiments show that the total cost of this model constructed in the study is 2323.94 RMB, which is less than 2623.35 RMB of the traditional path planning model, and the carbon emission is 85 kg less. It proves that the path optimization model constructed by the research has high economic and environmental benefits, and the designed and planned path is more in line with the needs of green economy than the traditional path, which has the potential to minimize carbon emissions during the fresh agricultural product cold chain distribution process while simultaneously promoting sustainable development.

The estimation of repair costs for car damage is a critical yet challenging task for insurance companies and repair shops. Accurate and the rapid predictions are essential for providing reliable cost estimates to customers. Traditional methods in this domain face multiple challenges, including manual processes and inaccuracies in repair cost estimation, as outlined in our article.

This paper introduces a novel approach that combines regression models with ontology reasoning to enhance the accuracy of car damage repair cost predictions. An Ontology for Car Damage (OCD)¹ ${}^{,}$² has been developed, which is meticulously structured and populated using Named Entity Recognition (NER) and Relation Extraction (RE) techniques. This ontology provides a comprehensive framework for organizing and understanding the complex domain of car damage, capturing essential semantic relationships and variables that significantly influence repair costs. By integrating OCD with seven regression models, such as Random Forest and Decision Tree, we have proposed a hybrid methodology that leverages both structured data and semantic understanding. Our approach not only accounts for typical variables such as the type and severity of damage, and labor costs but also identifies novel features through the use of SWRL (Semantic Web Rule Language) rules, enhancing the model’s predictive capabilities.

The performance of our models was evaluated using a substantial real-world dataset comprising over 300,000 records. This evaluation used metrics such as mean absolute error (MAE), root mean squared error (RMSE), and R-squared. The results indicate that our hybrid approach, which incorporates ontology reasoning, significantly outperforms traditional regression models.

The Random Forest model, especially when combined with the OCD ontology, showcased superior performance, exhibiting a minimal average deviation from the actual repair costs and achieving a low MAE.

This study’s findings demonstrate the potential of combining ontology reasoning with machine learning techniques for precise cost prediction in the automotive repair industry. Our methodology offers a robust tool for insurance companies and repair shops to generate more accurate, reliable, and automated cost estimates, ultimately benefiting both businesses and customers.

In multimodal sentiment analysis, achieving effective fusion among text, acoustic, and visual modalities for enhanced sentiment prediction is a crucial research topic. Recent studies typically employ tensor-based or attention-based mechanisms for multimodal fusion. However, the former fails to achieve satisfactory prediction performance, and the latter complicates the computation of fusion between non-textual modalities. Therefore, this paper proposes the multimodal sentiment analysis model based on Adaptive Feature Normalization and Attention Gating mechanism (AdaFN-AG). Firstly, facing highly synchronized non-textual modalities, we design the Adaptive Feature Normalization (AdaFN) method, which focuses more on sentiment features interaction rather than timing features association. In AdaFN, acoustic and visual modality features achieve cross-modal interaction through normalization, inverse normalization, and mix-up operations, with weights utilized for adaptive strength regulation of the cross-modal interaction. Meanwhile, we design the Attention Gating mechanism that facilitates cross-modal interactions between textual and non-textual modalities through cross-attention and captures timing associations, while the gating module concurrently regulates the intensity of these interactions. Additionally, we employ self-attention to capture the intrinsic correlations within single-modal features. Subsequently, we conduct experiments on three benchmark datasets for multimodal sentiment analysis, with the results indicating that AdaFN-AG outperforms the baselines across the majority of evaluation metrics. Through research and experiments, we validate that AdaFN-AG not only enhances performance by adopting appropriate methods for different types of cross-modal interactions while conserving computational resources but also verifies the generalization capability of the AdaFN method.

The accurate classification of endoscopic images is a challenging yet critical task in medical diagnostics, which directly affects the treatment and management of Gastrointestinal diseases. Misclassification can lead to incorrect treatment plans, adversely affecting patient outcomes. To address this challenge, our research aimed to develop a reliable computational model to improve the accuracy of classifying conditions of esophagitis and polyps. We focused on a subset of the Kvasir v1 secondary dataset, comprising 2000 endoscopic images evenly distributed across two classes: esophagitis and polyp. The goal was to leverage the strengths of both Machine Learning(ML) and Deep Learning(DL) to create a model that not only predicts with high accuracy but also integrates seamlessly into clinical workflows. To this end, we introduced a novel VRG-based ensemble image feature extraction technique, combining the powers of VGG, RF, and GB models to synthesize a robust feature set conducive to high-precision classification. The ensemble approach demonstrated a best-in-class performance with the GB model achieving an outstanding 99.73% accuracy in detecting esophagitis and polyps. The practical implications of these results are substantial, indicating that our method can significantly improve diagnostic accuracy in real-world settings, reduce the rate of misdiagnosis, and contribute to the efficient and effective treatment of patients, ultimately enhancing the quality of healthcare services. With the successful application of our proposed method to a controlled dataset, future work involves deploying the model in clinical environments and expanding its application to a broader spectrum of Gastrointestinal conditions across multi-class datasets.

Contamination of harmful additives in fruits has become a concerning norm these days. Owing to the great popularity of fruits, dishonest vendors frequently use harmful chemicals to contaminate fruits to extend their shelf life, which is extremely dangerous for the general public's health. To mitigate this issue, machine-learning algorithms like Decision Tree Classifier, Naïve Bayes and a deep learning model named “DurbeenNet” are evaluated separately. Alongside, a computer vision-based detection method coupled with a hybrid model is proposed that combines deep learning and chemical sensor. Formaldehyde Detection Sensor is used in this experiment to take reading of the sensor data. Mango, Apple, Banana, and Malta are taken as sample fruits in this study. Sensor data for both fresh and chemical-mixed fruit is newly collected using Formaldehyde Detection Sensor. The above mentioned sensor data along with the previously captures images of both fresh and chemical-mixed state are being integrated to a hybrid model. Among two machine learning algorithms naïve bayes come up with 82 % accuracy. Using both sensor data and captured image data, the proposed model “SensorNet” provides highest accuracy of 97.03 % which is substantial than “DurbeenNet” model's accuracy. Through the utilization of these fruit samples, formaldehyde detection sensor provides instantaneous detection, identifying the specific toxic substances present in the contaminated fruits.

Detecting descending stairs and floors is a crucial aspect of implementing autonomous systems in smart wheelchairs. When the obstacle detection system used in wheelchairs fails to accurately identify descending stairs, it can lead to severe consequences for users, including injuries or, in the worst-case scenario, fatal accidents. Therefore, there is a pressing need for an algorithm that not only exhibits high accuracy in detecting obstacles on descending stairs but also operates with minimal computational delay to ensure an immediate response in wheelchair braking. In this research, We utilize the GLCM technique to extract texture characteristics. Out of these methods, the Decision Tree exhibits the highest accuracy, reaching 94%, with a remarkably fast computational time of 0.01299 s. These promising results were achieved by utilizing the GLCM method with a distance of 2 and an angle of 45°. The accuracy obtained has increased by 2.5% compared to the previous research. Such a level of accuracy, coupled with fast computational performance, enables smart wheelchairs to effectively assist users in identifying obstacles while descending stairs.