Search

Mild cognitive impairment (MCI) is a transitional stage from normal aging to Alzheimer’s disease (AD). Early detection of MCI is essential, as it offers a last opportunity for interventions to slow or prevent progression to AD. However, identifying effective biomarkers for screening remains challenging. While declines in perception and action often precede visible neurodegenerative changes, studies on visual pathway biomarkers for MCI detection are limited. In this study, we focused on electroencephalography with steady-state visual evoked potentials (EEG-SSVEP), known for its high-resolution, real-time monitoring of brain response to flicker stimulation, as a promising method for early MCI identification. We collected EEG-SSVEP data from 24 healthy controls and 25 MCI patients, extracting 166 EEG-SSVEP biomarkers, including lobe power ratio, lobe connectivity ratio, and band connectivity ratio, to assess the visual pathway’s dorsal and ventral streams related to cognitive decline. By employing a biomarker selection method, we identified six key EEG-SSVEP biomarkers as the most relevant for distinguishing between healthy controls and MCI patients. Subsequently, these six biomarkers were utilized to train a support vector machine for early detection of MCI. The results showed an accuracy rate of 95.69%, a sensitivity of 92.28%, and a specificity of 95.58%. This study offers valuable insights into enhancing the early detection of MCI by leveraging EEG-SSVEP data and machine learning to assess cognitive decline within the dorsal and ventral streams of the brain.

Human hair segmentation is essential for face recognition and for achieving natural transformation of style transfer. However, it remains a challenging task due to the diverse appearances and complex patterns of hair in image. In this study, we propose a novel method utilizing diffusion-based generative models, which have been extensively researched in recent times, to effectively capture and to finely segment human hair. In diffusion-based models, an internal visual representation during the denoising process contains pixel-level rich information. Inspired by this aspect, we introduce diffusion-based models for segmenting fine-grained human hair. Specifically, we extract the representation from the diffusion-based models, which contains pixel-level semantic information, and then train a segmentation network using it. Particularly, to more finely segment human hair, our approach employs the representation from a text-to-image diffusion model, conditioned on text information, to extract more relevant information for human hair, thereby predicting detailed hair masks. To validate our method, we conducted experiments on three distinct hair-related datasets with unique characteristics: Figaro-1k, CelebAMask-HQ, and Face Synthetics. The experimental results show the improved performance of our proposed method across all three datasets, outperforming existing methods in terms of mIoU (mean intersection over union), accuracy, precision, and F1-score. This is particularly evident in its ability to accurately capture and finely segment human hair from background and non-hair elements. This demonstrates the effectiveness of our method in accurately and finely segmenting human hair with complex characteristics. Our research contributes not only to the fine-grained segmentation of human hair but also to the application of generative models in semantic segmentation tasks. We hope that the proposed method will be applied for detailed semantic segmentation in various fields in the future.

Decision tree algorithm (DT) is a commonly used data mining method for classification and regression. DT repeatedly divides a dataset into pure subsets based on impurity measurements such as entropy and Gini. Then relatively “pure” partitions consisting of observations with the (almost) same class are obtained. Gini index is one of the representative indices for measuring the impurity of data. However, the Gini index does not take into account distances between classes. If the distances between classes are considered when measuring impurity, the decision tree algorithm can distinguish clearly observations with different classes. To the end, a new decision tree algorithm based on Rao-Stirling index is proposed considering distances between classes. Rao-Stirling index considers distances between classes in such a way that weights more to pairs of references in more distant classes when measuring data impurity. Experimental results indicate that the proposed method is superior in terms of accuracy, implying that considering the distances between classes can help improve accuracy in DT.

JPEG compression is one of the major image compression methods and is widely used on the Internet. In addition, identifying traces of JPEG compression and double JPEG compression (DJPEG) is crucial in the image forensics field. Therefore, JPEG compression detection and DJPEG compression detection are two of the popular image authentication methods. Many feature-based JPEG detection methods have been proposed for that purpose, and there have been outstanding improvements in DJPEG detection with the development of deep learning. A number of anti-forensics of JPEG detection that counter feature-based detectors have been proposed but only a few techniques that counter DJPEG have been researched. This paper explores whether JPEG reconstruction methods, including restoration and anti-forensics of JPEG detection, can deceive JPEG and DJPEG detectors. We demonstrate that existing anti-forensics of JPEG detection can deceive both JPEG and DJPEG detectors well but perform poorly in non-aligned cases and degrade the image quality. We propose a convolutional neural network (CNN) based anti-forensics method to improve the performance of anti-forensics so that they can proficiently deceive JPEG and DJPEG detectors with higher image quality. Moreover, we explore the generalization algorithm to handle the real scenario.

Deep neural networks (DNNs) have been making progress in many ways. DNNs are typically used to model complex nonlinearity of high-dimensional data in regression or classification problems. As DNNs contain additional hidden layers, they generally improve performance but increase the number of parameters to train, thereby extending the learning time. Many studies, such as those employing Dropout and regularization methods, have undertaken to solve these problems. The method proposed in this paper is an additive ensemble neural networks (AENNs), by which a boosting mechanism of an ensemble methodology is applied to the neural networks instead of regularization techniques. That is, the model by AENNs is obtained by sequentially combining several simple shallow network models. Experiments showed that AENNs yield better results than conventional DNNs and machine learning methods for regression and classification problems, thereby alleviating the troublesome model complexity issue.

In this paper, we propose element-wise adaptive threshold methods for learned iterative shrinkage thresholding algorithms. The threshold for each element is adapted in such a way that it is set to be smaller when the previously recovered estimate or the current one-step gradient descent at that element has a larger value. This adaptive threshold gives a lower misdetection probability of the true support, which speedups the convergence to the optimal solution. We show that the proposed element-wise threshold adaption method has better convergence rate than the existing non-adaptive threshold methods. Numerical results show that the proposed neural network has the best recovery performance among the tested algorithms. In addition, it is robust to the sparsity mismatch, which is very desirable in the case of unknown signal sparsity.

As IoT devices are always connected to mobile devices or other computing devices via the Internet, clever malwares targeting IoT devices or other computing devices connected to IoT devices are emerging. Therefore, effective IoT security research is needed to respond to hacking attacks by these kinds of malware. This paper studied the method of identifying and analyzing malware combined with social engineering from the perspective of digital forensics. The paper classified and analyzed intelligent malware characteristics and proposed a method of quickly identifying and analyzing the malware that secretly intruded into the devices installed with Android, Linux OS, using digital forensics techniques. Moreover, this paper proved its effectiveness by applying this investigation method to two actual malware cases. The research outcomes will be useful in responding to increasingly clever malware attacking IoT devices.