Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder considered by challenges in communication, social interaction, and behavior. Early and accurate detection remains a challenge due to subjective clinical ...
详细信息
ISBN:
(数字)9798331537555
ISBN:
(纸本)9798331537562
Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder considered by challenges in communication, social interaction, and behavior. Early and accurate detection remains a challenge due to subjective clinical assessments. This work presents an innovative hybrid framework that incorporates sentiment analysis from social media data with deep learning techniques applied to clinical data to enhance ASD prediction. Leveraging Twitter sentiment, a Long Short-Term Memory (LSTM) model captures public emotional and behavioral patterns, while clinical data undergoes deep learning-based character-level embedding using a Convolutional Neural Network (CNN) with transfer learning. The fusion of these models through a weighted sum mechanism results in a comprehensive ASD detection system. Experimental outcomes demonstrate an accuracy of 99.3% with a loss of 1.45%, significantly outperforming traditional diagnostic approaches. Visual analyses further validate key correlations between assessment scores and ASD diagnoses. Despite limitations such as misinformation in social media and the need for larger clinical datasets, the proposed approach highlights the potential of integrating social and clinical perspectives for improved ASD diagnosis. Future work will explore broader social media sources and advanced deep learning architectures to refine the model, ultimately contributing to early intervention and progressed results for individuals with ASD.
Multi-disease conditions strain the body’s defenses, complicating recovery and increasing mortality risk. Therefore, effective concurrent prevention of multiple diseases is essential for mitigating complications and ...
详细信息
Self-supervised learning has shown outstanding performance on speaker verification, and the 2-stage frameworks have more comprehensive training schemes, which typically exhibit better performance. They utilize cluster...
详细信息
ISBN:
(数字)9798350368741
ISBN:
(纸本)9798350368758
Self-supervised learning has shown outstanding performance on speaker verification, and the 2-stage frameworks have more comprehensive training schemes, which typically exhibit better performance. They utilize clustering to obtain pseudo-labels, which are then used as the supervision signal in stage 2. However, these pseudo-labels often contain a significant amount of noisy labels, severely impacting speaker verification performance. In this paper, we propose a dynamic self-supervised pseudo-label correction method based on batch-scale training. By filtering and correcting samples based on the loss and prediction distribution, our method better aligns with the dynamic training process and achieves EER(%) of 1.33, 1.56 and 2.78 on the test sets of Voxceleb-O, E, H.
Human action recognition (HAR) is a crucial field in computer vision with applications ranging from video surveillance to human-computer interaction. This study explores an efficient framework for HAR by leveraging hu...
详细信息
Classifying tiny objects in remote sensing images (e.g., a 20x20 pixels target within a resolution 1000x1000 image) is a significant challenge. This paper adopts a fused FPN (Feature Pyramid Network) to enhance the fe...
详细信息
ISBN:
(数字)9798331521165
ISBN:
(纸本)9798331521172
Classifying tiny objects in remote sensing images (e.g., a 20x20 pixels target within a resolution 1000x1000 image) is a significant challenge. This paper adopts a fused FPN (Feature Pyramid Network) to enhance the feature extraction capability for tiny objects. Additionally, the proposed multi-scale crop method can more effectively focus on the target objects. This network architecture improves overall accuracy by 14.4%.
Owing to the rapid evolution of technologies and project requirements, organizations need to upgrade the code base in their software projects to a new version of the programming language or even translating to an enti...
详细信息
With the increasing need for scalable, energy-efficient sensing solutions in smart cities and buildings, traditional temperature monitoring systems face limitations due to higher manufacturing costs and rigid designs....
详细信息
ISBN:
(数字)9798331518424
ISBN:
(纸本)9798331518431
With the increasing need for scalable, energy-efficient sensing solutions in smart cities and buildings, traditional temperature monitoring systems face limitations due to higher manufacturing costs and rigid designs. Printed fabrication offers a flexible and economical alternative that enhances potential for volume deployment, ubiquitous integration and the increased spatial resolution of monitoring. In this work, we present a novel environmental temperature monitoring system that leverages printed fabrication for antenna and sensor for such applications. A high measured sensitivity (i.e. 3.1% Temperature coefficient of resistance) is shown for the designed printed temperature sensor. A measured 93MHz bandwidth coverage is shown for the designed printed, flexible dual-patch array WiFi antenna.
The research focuses on developing an electroencephalography (EEG) based emotion recognition system to identify happy, neutral, and negative emotions. The suggested framework uses Simple Recurrent Neural Networks (Sim...
详细信息
An M-sequence generated by a primitive polynomial has many interesting and desirable properties. A pseudorandom array is the two-dimensional generalization of an M-sequence. Similarly to primitive polynomials, there a...
详细信息
Unsupervised anomalous sound detection aims to detect unknown anomalous sounds by training a model using only normal audio data. Despite advancements in self-supervised methods, the issue of frequent false alarms when...
详细信息
ISBN:
(数字)9798350368741
ISBN:
(纸本)9798350368758
Unsupervised anomalous sound detection aims to detect unknown anomalous sounds by training a model using only normal audio data. Despite advancements in self-supervised methods, the issue of frequent false alarms when handling samples of the same type from different machines remains unresolved. This paper introduces a novel training technique called one-stage supervised contrastive learning (OS-SCL), which significantly addresses this problem by perturbing features in the embedding space and employing a one-stage noisy supervised contrastive learning approach. On the DCASE 2020 Challenge Task 2, it achieved 94.64% AUC, 88.42% pAUC, and 89.24% mAUC using only Log-Mel features. Additionally, a time-frequency feature named TFgram is proposed, which is extracted from raw audio. This feature effectively captures critical information for anomalous sound detection, ultimately achieving 95.71% AUC, 90.23% pAUC, and 91.23% mAUC. The source code is available at: ***/huangswt/OS-SCL.
暂无评论