Eating behavior is highly heterogeneous across individuals and cannot be fully explained using only the degree of obesity. We utilized unsupervised machine learning and functional connectivity measures to explore the ...
详细信息
Eating behavior is highly heterogeneous across individuals and cannot be fully explained using only the degree of obesity. We utilized unsupervised machine learning and functional connectivity measures to explore the heterogeneity of eating behaviors measured by a self-assessment instrument using 424 healthy adults (mean +/- standard deviation [SD] age = 47.07 +/- 18.89 years;67% female). We generated low-dimensional representations of functional connectivity using resting-state functional magnetic resonance imaging and estimated latent features using the feature representation capabilities of an autoencoder by nonlinearly compressing the functional connectivity information. The clustering approaches applied to latent features identified three distinct subgroups. The subgroups exhibited different levels of hunger traits, while their body mass indices were comparable. The results were replicated in an independent dataset consisting of 212 participants (mean +/- SD age = 38.97 +/- 19.80 years;35% female). The model interpretation technique of integrated gradients revealed that the between-group differences in the integrated gradient maps were associated with functional reorganization in heteromodal association and limbic cortices and reward-related subcortical structures such as the accumbens, amygdala, and caudate. The cognitive decoding analysis revealed that these systems are associated with reward- and emotion-related systems. Our findings provide insights into the macroscopic brain organization of eating behavior-related subgroups independent of obesity. We systematically investigated the heterogeneity of eating behavior traits of healthy adults using unsupervised machine learning and functional connectivity. We identified three distinct subgroups showing different eating behavior traits independent of the degree of ***
Sensors, wearables, mobile devices, and other Internet of Things (IoT) devices are becoming increasingly integrated into all aspects of our lives. They are capable of gathering enormous amounts of data, such as image ...
详细信息
Sensors, wearables, mobile devices, and other Internet of Things (IoT) devices are becoming increasingly integrated into all aspects of our lives. They are capable of gathering enormous amounts of data, such as image data, which can then be sent to the cloud for processing. However, this results in an increase in network traffic and latency. To overcome these difficulties, edge computing has been proposed as a paradigm for computing that brings processing closer to the location where data is produced. This paper explores the merging of cloud and edge computing for IoT and investigates approaches using machine learning for dimensionality reduction of images on the edge, employing the autoencoder deep learning-based approach and principal component analysis (PCA). The encoded data is then sent to the cloud server, where it is used directly for any machine learning task without significantly impacting the accuracy of the data processed in the cloud. The proposed approach has been evaluated on an object detection task using a set of 4000 images randomly chosen from three datasets: COCO, human detection, and HDA datasets. Results show that a 77% reduction in data did not have a significant impact on the object detection task's accuracy.
Background: High-throughput methodologies such as microarrays and next-generation sequencing are routinely used in cancer research, generating complex data at different omics layers. The effective integration of omics...
详细信息
Background: High-throughput methodologies such as microarrays and next-generation sequencing are routinely used in cancer research, generating complex data at different omics layers. The effective integration of omics data could provide a broader insight into the mechanisms of cancer biology, helping researchers and clinicians to develop personalized therapies. Results: In the context of CAMDA 2017 Neuroblastoma Data Integration challenge, we explore the use of Integrative Network Fusion (INF), a bioinformatics framework combining a similarity network fusion with machine learning for the integration of multiple omics data. We apply the INF framework for the prediction of neuroblastoma patient outcome, integrating RNA-Seq, microarray and array comparative genomic hybridization data. We additionally explore the use of autoencoders as a method to integrate microarray expression and copy number data. Conclusions: The INF method is effective for the integration of multiple data sources providing compact feature signatures for patient classification with performances comparable to other methods. Latent space representation of the integrated data provided by the autoencoder approach gives promising results, both by improving classification on survival endpoints and by providing means to discover two groups of patients characterized by distinct overall survival (OS) curves.
Mostvexisting identification and tackling of chaos in swarm drone missions focus on single drone scenarios. There is a need to assess the status of a system with multiple drones, hence, this research presents an on-th...
详细信息
Mostvexisting identification and tackling of chaos in swarm drone missions focus on single drone scenarios. There is a need to assess the status of a system with multiple drones, hence, this research presents an on-the-fly chaotic behavior detection model for large numbers of flying drones using machine learning techniques. A succession of three Artificial Intelligence knowledge discovery procedures, Logistic Regression (LR), Convolutional Neural Network (CNN), Gaussian Mixture Models (GMMs) and Expectation-Maximization (EM) were employed to reduce the dimension of the actual data of the swarm of drone's flight and classify it as non-chaotic and chaotic. A one-dimensional, multi-layer perceptive, deep neural network-based classification system was also used to collect the related characteristics and distinguish between chaotic and non-chaotic conditions. The Rossler system was then employed to deal with such chaotic conditions. Validation of the proposed chaotic detection and mitigation technique was performed using real-world flight test data, demonstrating its viability for real-time implementation. The results demonstrated that swarm mobility horizon-based monitoring is a viable solution for real-time monitoring of a system's chaos with a significantly reduced commotion effect. The proposed technique has been tested to improve the performance of fully autonomous drone swarm flights.
In November 2019, the coronavirus disease outbreak began, caused by the novel severe acute respiratory syndrome coronavirus 2. In just over two months, the unprecedented rapid spread resulted in more than 10,000 confi...
详细信息
In November 2019, the coronavirus disease outbreak began, caused by the novel severe acute respiratory syndrome coronavirus 2. In just over two months, the unprecedented rapid spread resulted in more than 10,000 confirmed cases worldwide. This study predicted the infectious spread of coronavirus disease in the contiguous United States using a convolutional autoencoder with long short-term memory and compared its predictive performance with that of the convolutional autoencoder without long short-term memory. The epidemic data were obtained from the World Health Organization and the US Centers for Disease Control and Prevention from January 1st to April 6th, 2020. We used data from the first 366,607 confirmed cases in the United States. In this study, the data from the Centers for Disease Control and Prevention were gridded by latitude and longitude and the grids were categorized into six epidemic levels based on the number of confirmed cases. The input of the convolutional autoencoder with long short-term memory was the distribution of confirmed cases 14 days before, whereas the output was the distribution of confirmed cases 7 days after the date of testing. The mean square error in this model was 1.664, the peak signal-to-noise ratio was 55.699, and the structural similarity index was 0.99, which were better than those of the corresponding results of the convolutional autoencoder. These results showed that the convolutional autoencoder with long short-term memory effectively and reliably predicted the spread of infectious disease in the contiguous United States.
Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder that can reduce quality of life and burden families. However, there is a lack of objectivity in clinical diagnosis, so it is very important to de...
详细信息
Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder that can reduce quality of life and burden families. However, there is a lack of objectivity in clinical diagnosis, so it is very important to develop a method for early and accurate diagnosis. Multi-site data increases sample size and statistical power, which is convenient for training deep learning models. However, heterogeneity between sites will affect ASD recognition. To solve this problem, we propose a multi-site anti-interference neural network for ASD classification. The resting state brain functional image data provided by the multi-site is used to train the ASD classification model. The model consists of three modules. First, the site feature extraction module is used to quantify the inter-site heterogeneity, in which the autoencoder is used to reduce the feature dimension. Secondly, the presentation learning module is used to extract classification features. Finally, the anti-interference classification module uses the output of the first two modules as labels and inputs for multi-task adversarial training to complete the representation learning that is not affected by the confounding of sites, so as to realize the adaptive anti-interference ASD classification. The results show that the average accuracy of ten-fold cross validation is 75.56%, which is better than the existing studies. The innovation of our proposed method lies in the problem that the traditional single-task deep learning ASD classification model will be affected by the heterogeneity of multi-site data and interfere with the classification. Our method eliminates the influence of multi-site factors on feature extraction through multi-task adversarial training, so that the model can better adapt to the heterogeneity of multi-site data. Meanwhile, large-scale 1DconV is introduced to extract features of brain functional network, which provides support for the interpretability of the model. This method is expected to take a
An intrusion detection system (IDS) is one of the most effective ways to secure a network and prevent unauthorized access and security attacks. But due to the lack of adequately labeled network traffic data, researche...
详细信息
An intrusion detection system (IDS) is one of the most effective ways to secure a network and prevent unauthorized access and security attacks. But due to the lack of adequately labeled network traffic data, researchers have proposed several feature representations models over the past three years. However, these models do not account for feature generalization errors when learning semantic similarity from the data distribution and may degrade the performance of the predictive IDS model. In order to improve the capabilities of IDS in the era of Big Data, there is a constant need to extract the most important features from large-scale and balanced network traffic data. This paper proposes a semi-supervised IDS model that leverages the power of untrained autoencoders to learn latent feature representations from a distribution of input data samples. Further, distance function-based clustering is used to find more compact code vectors to capture the semantic similarity between learned feature sets to minimize reconstruction loss. The proposed scheme provides an optimal feature vector and reduces the dimensionality of features, reducing memory requirements significantly. Multiple test cases on the IoT dataset MQTTIOT2020 are conducted to demonstrate the potential of the proposed model. Supervised machine learning classifiers are implemented using a proposed feature representation mechanism and are compared with shallow classifiers. Finally, the comparative evaluation confirms the efficacy of the proposed model with low false positive rates, indicating that the proposed feature representation scheme positively impacts IDS performance.
This research uses deep learning to address the high peak-toaverage power ratio (PAPR) in orthogonal frequency division multiplexing (OFDM), which is critical for wireless communications. Although a PAPRreducing netwo...
详细信息
This research uses deep learning to address the high peak-toaverage power ratio (PAPR) in orthogonal frequency division multiplexing (OFDM), which is critical for wireless communications. Although a PAPRreducing network (PRNet), which is a deep learning model, can be used to suppress the PAPR, its computational cost is huge. In this research, the number of layers in a PRNet model is optimized and a fully connected layer is replaced with a convolution layer to reduce the computational load.
The usage of credit card has increased dramatically due to a rapid development of credit cards. Consequently, credit card fraud and the loss to the credit card owners and credit cards companies have been increased dra...
详细信息
The usage of credit card has increased dramatically due to a rapid development of credit cards. Consequently, credit card fraud and the loss to the credit card owners and credit cards companies have been increased dramatically. Credit card Supervised learning has been widely used to detect anomaly in credit card transaction records based on the assumption that the pattern of a fraud would depend on the past transaction. However, unsupervised learning does not ignore the fact that the fraudsters could change their approaches based on customers' behaviors and patterns. In this study, three unsupervised methods were presented including autoencoder, one-class support vector machine, and robust Mahalanobis outlier detection. The dataset used in this study is based on real-life data of credit card transaction. Due to the availability of the response, fraud labels, after training the models the performance of each model was evaluated. The performance of these three methods is discussed extensively in the manuscript. For one-class SVM and auto encoder, the normal transaction labels were used for training. However, the advantages of robust Mahalanobis method over these methods is that it does not need any label for its training.
暂无评论