Haptic simulation systems use the sensation of touch to allow users to experience virtual surroundings. In interfaces that are based on impedance, haptics controllers are sampled data that use angular position and vel...
详细信息
In the pursuit of sustainable living and enhanced user comfort, the integration of smart technologies into everyday devices has become paramount. This paper highlights the importance of dynamic lighting control in imp...
详细信息
ISBN:
(数字)9798350353778
ISBN:
(纸本)9798350353785
In the pursuit of sustainable living and enhanced user comfort, the integration of smart technologies into everyday devices has become paramount. This paper highlights the importance of dynamic lighting control in improving energy efficiency and the overall quality of life. Traditional lighting systems often fail to account for the dynamic nature of human activities, resulting in unnecessary energy wastage and suboptimal lighting conditions. The proposed system seeks to bridge the gap between existing lighting systems and the evolving needs of modern indoor spaces, with its focus on adaptability, efficiency, and user-centricity, promoting a more comfortable and nurturing environment. Although earlier studies have explored various aspects of smart lighting technologies, there remains a paucity of comprehensive solutions that integrate machine learning with sensor technologies to create truly adaptive environments. Novel algorithms have been utilized to optimize both visual comfort and energy efficiency. This system achieves up to 45-70% energy savings compared to existing lighting systems, making it a promising solution for both environmental and economic concerns.
Sleep apnea syndrome (SAS) is common but people are generally unaware of this condition, as it occurs during sleep. A method was previously proposed to estimate respiration during sleep in a non-contact and non-constr...
详细信息
ISBN:
(数字)9798350355079
ISBN:
(纸本)9798350355086
Sleep apnea syndrome (SAS) is common but people are generally unaware of this condition, as it occurs during sleep. A method was previously proposed to estimate respiration during sleep in a non-contact and non-constraint manner using a dept. camera that is suitable for everyday use. The method was validated on subjects who were awake but remains untested under sleeping conditions. In this study, 12 men aged 65 years or older were tested under actual sleeping conditions for eight days. The effectiveness and accuracy of the proposed method were confirmed through a comparison with commercially available products.
Affective state plays a key role in emotion recognition, influencing various applications from mental health evaluation to human-machine interaction. These states can be described using two primary aspects: valence, w...
详细信息
ISBN:
(数字)9798331540067
ISBN:
(纸本)9798331540074
Affective state plays a key role in emotion recognition, influencing various applications from mental health evaluation to human-machine interaction. These states can be described using two primary aspects: valence, which indicates the pleasantness of the speech, and arousal, which relates to the intensity of the emotion. This study aims to automate the prediction of valence and arousal values from speech data, addressing the limitations of manual annotation and human error. The IIITSAINT-EMOMDB dataset, created from Malayalam films, uses the Circumplex Model of Affect to map emotions such as happy, sad, angry, and calm. The approach leverages machine learning models, including Support Vector Regression (SVR), Artificial Neural Networks (ANN), and Convolutional Neural Networks (CNN), for prediction. Experimental results demonstrate that the traditional models like SVR and ANN are outperformed by the CNN model, which achieved a mean squared error (MSE) of 0.3 for valence and 0.1 for arousal. These findings highlight the power of automated emotion recognition in large-scale audio datasets, offering promising applications in affective computing and psychological analysis.
Drowsiness during driving includes high risks for the driver, the co-passengers, and the other people on the road. Advance Driver Assistance Systems (ADAS) have been proposed to reduce these risks. This work introduce...
详细信息
Cancer is the leading cause of premature mortality. Globally, the prevalence of cancer is on the rise, and early-stage diagnosis is crucial for cancer recovery and survival. However, the presence of circulating tumor ...
详细信息
ISBN:
(数字)9798331531195
ISBN:
(纸本)9798331531201
Cancer is the leading cause of premature mortality. Globally, the prevalence of cancer is on the rise, and early-stage diagnosis is crucial for cancer recovery and survival. However, the presence of circulating tumor cells (CTCs) and heterogeneity within the tumor makes it hard to track how cancer changes over time and how resistance to treatment develops. Unfortunately, the cancer diagnosis precision remains inadequate, endangering precious human lives and necessitating costly procedures for accurate treatment. Machine learning (ML) has proven capable of discovering meaningful information and making predictions, particularly with large datasets. Moreover, quantum computers (QCs) inherently possess superposition and parallel computation, features not found in classical computing. Researchers have demonstrated that quantum neural networks (QNNs) achieve a significantly better effective dimension than classical NNs. In this context, this study investigates the application of classical and quantum machine learning (QML) algorithms to classify breast cancer images. Region-based segmentation is used to identify regions of interest and image processing algorithms are performed on segmented breast cancer images to extract meaningful features (e.g., perimeter, area, compactness). An user interface (UI) based software is created to swiftly distinguish benign and malignant instances during image analysis. Although QNNs perform poorly than the classical NNs, there is still a long way to go in developing feature maps and techniques for better performance. In contrast, the classical NNs have gone a long way with well-established algorithms and packages available to us. This research shows that with further development, QML has the potential to enhance cancer diagnosis accuracy and speed, ultimately improving early detection and treatment outcomes.
This contribution describes new useful geometric transformations using the tensor product. The geometric transformations are used widely in many applications, especially in CAD/CAM systems, systems for Civil Engineeri...
详细信息
Audio signals being high-dimensional and complex pose challenges for classical machine learning techniques in terms of computation and generalization on real-world data. This paper evaluates the use of hybrid quantum-...
详细信息
ISBN:
(数字)9798350370249
ISBN:
(纸本)9798350370270
Audio signals being high-dimensional and complex pose challenges for classical machine learning techniques in terms of computation and generalization on real-world data. This paper evaluates the use of hybrid quantum-classical convolutional neural networks (QC-CNN) that leverage quantum effects like superposition and entanglement for audio classification using mel-spectrograms obtained from audio data. Evaluated on both small-sized and large-sized datasets, the proposed QC-CNN model gave comparable training accuracy with classical CNN (Convolutional Neural Network) on the smaller dataset but outperformed classical CNN on test accuracy ($\mathbf{95.04 \%}$ vs $\mathbf{92.88 \%}$) for a larger birdsong dataset and reduced overfitting, thus highlighting the potential advantages of QC-CNNs for audio data. The QC-CNN exhibited higher cross-entropy loss in case of the small-sized dataset which was further significantly reduced when evaluated on the large-sized birdsong dataset. The work demonstrates the application of QC-CNN for audio classification.
The advancement of automation technologies made lives simpler and easier in all aspects. In today's world, automated systems [1] are replacing manual systems progressively. In this paper, an overview of current an...
详细信息
Cataract is a common eye condition that causes clouding of the eye's natural lens, resulting in blurry vision and it is very common in older ages people. However, it can also occur in younger aged people due to va...
Cataract is a common eye condition that causes clouding of the eye's natural lens, resulting in blurry vision and it is very common in older ages people. However, it can also occur in younger aged people due to various genetical effects, diabetes, smoking, and prolonged exposure to sunlight. Cataracts can be treated through surgery to remove the clouded lens and replaced by an artificial one, which can significantly improve the vision. This problem can be reduced through early detection, regular eye examinations etc. In this research, a Contrast Limited Histogram Equalization (CLAHE) of retinal fundus images is applied for a better presentation of the above cataract effects. These improved images are fed to a machine learning-based model for effective detection. Finally, the ensembling of the proposed machine learning-based classifiers is performed for the accurate detection of cataracts. The proposed model was tested experimentally on a real dataset and achieves excellent performance with accuracy, precision, and recall scores of 99.67%, 100%, and 99.25%, respectively which outperform the baseline methods. The robustness of the proposed ensemble model is evaluated through 5-fold cross-validation and achieved an average accuracy of 99.6%. The implementation code can be accessed from the GitHub repository: https://***/nahid-tech/cataract-detection.
暂无评论