Cognitive Computation and Systems最新文献

Mental health disorders have a global prevalence of 25%, according to the WHO, and this is exacerbated by factors such as stigma, geographical location, and a worldwide shortage of practitioners. Mental health chatbots have been developed to address these barriers, but these systems lack key features such as emotion recognition, personalisation, multilingual support, and ethical appropriateness. This paper introduces an innovative mental health support system that integrates BERT-based emotional distress detection with a psychological first aid (PFA)-generative pre-trained transformer (PFA-GPT) model, providing an emotion-aware PFA chatbot. The methodology leverages deep learning models, utilising bidirectional encoder representations from transformers (BERT) for emotional distress detection and fine-tuning GPT-3.5 on therapy transcripts for PFA chatbot development. The findings demonstrate BERT's superior accuracy (93%) for emotional distress detection compared to bidirectional long short-term memory. The multilingual PFA chatbot developed using the PFA-GPT model demonstrated superior BERT scores (exceeding 83%) and proficiently provided ethical PFA. A proof of concept has been developed to illustrate the integration of the emotional distress detection model with the novel generative conversational agent for PFA. This integrated approach holds significant potential in overcoming existing barriers to mental health support and has the potential to transform mental health support, offering timely and accessible care through AI-powered psychological interventions.

Brain networks provided powerful tools for the analysis and diagnosis of epilepsy. This paper performed a pairwise comparative analysis on the brain networks of Benign Childhood Epilepsy with Centrotemporal Spikes (BECTS): spike group (spike), non-spike group (non-spike), and control group (control). In this study, fragments with and without interictal spikes in electroencephalograms of 13 BECTS children during non-rapid eye movement sleep stage I (NREMI) were selected to construct dynamic brain function networks to explore the functional connectivity (FC). Graph theory and statistical analysis were exploited to investigate changes in FC across different brain regions in different frequency bands. From this study, we can draw the following conclusions: (1) Both spike and non-spike have lower energy in each brain region on the γ band. (2) With the increase of the frequency band, the FC strength of spike, non-spike and control groups are all weakened. (3) Spikes are correlated with brain network efficiency and the small-world property. (4) Spikes increase the FC of temporal, parietal and occipital regions except in the γ band and the absence of spikes weakens the FC of the entire brain region.

Health insurance plays a vital role during medical emergencies in the coverage against medical expenses. Insurance fraud is an international challenge that affects most economies worldwide. Government and private companies offer many insurance schemes. The successful implementation of numerous health insurance programs offered for the public by and large are often threatened by corruption, fraud, and numerous other data-related issues. Further the procedure for acclaiming the insurance money is not only critical in terms of verification of claims but tedious and time consuming also. To help redress these problems, blockchain technology can be utilised as is it offers improved security, transparency, auditability, privacy, accountability along with many other advantages. The goal is to create and implement a blockchain-based solution for efficient functioning of insurance system and to prevent such health insurance systems from going bankrupt. The authors have proposed an insurance claim model, MedBlockSure using blockchain architecture for creating interoperability between the insurer, the hospital and the insurance company. The model will aid in maintaining transparency between the insurer and the company while eliminating the requirement of middlemen or agents. The conceptual view of the proposed system using sequence and use case diagrams and data management framework and smart claim processing system is demonstrated.

Low-light object detection is needed for ensuring security, enabling surveillance, and enhancing safety in diverse applications, including autonomous vehicles, surveillance systems, and search and rescue operations. A comprehensive study on low-light object detection is presented using state-of-the-art you only look once (YOLO) models, including YOLOv3, YOLOv5, YOLOv6, and YOLOv8, aiming to enhance detection performance under challenging low-light conditions. The ExDark dataset is a dataset that consists of adequate low-light images, modified to simulate realistic low-light scenarios, and employed for evaluation. The deep learning algorithm optimises YOLO's architecture for low-light detection by adapting the network structure and training strategies while preserving the algorithm's integrity. The experimental results show that YOLOv8 consistently outperforms baseline models, achieving significant improvements in accuracy and robustness in low-light scenarios. The deep learning algorithm that acquired the best score, YOLOv8s, had a mean average precision score of 0.5513. This work contributes to the field of low-light object detection, offering promising solutions for real-world applications like nighttime surveillance and autonomous navigation in low-light conditions, addressing the growing demand for advanced low-light object detection.

The firing of spatial cells in the entorhinal-hippocampal structure is believed to enable the formation of a cognitive map for the environment. Inspired by the spatial cognitive mechanism of the rat's brain, the authors proposed a real-time cognitive map construction method based on the entorhinal-hippocampal working mechanism. Firstly, based on the physiological properties of the rat's brain, the authors constructed an entorhinal-hippocampal CA3 neurocomputational model for path integration. Then, the transformation relationship between the cell plate and the real environment is used to solve the robot's position. Path integration inevitably generates cumulative errors, which require loop-closure detection and pose optimisation to eliminate errors. To solve the problem that the RatSLAM algorithm is slow in pose optimisation, the authors proposed a pose optimisation method based on a multi-layer CA1 place cell to improve the speed of pose optimisation. To validate the method, the authors designed simulation experiments, dataset experiments, and physical experiments. The experimental results showed that compared to other brain-like SLAM algorithms, the authors’ method possesses outstanding performance in path integration accuracy and map construction speed. As a result, the authors’ method can endow mobile robots with the ability to quickly and accurately construct cognitive maps in complex and unknown environments.

Mixed sentiment classification (MSC) technology has a significant research value and application potential in understanding and analysing sentimental interactions. In the process of identifying and analysing complex sentiments, it is still necessary to overcome the difficulties of multi-dimensional sentiment recognition and improve sensitivity to subtle sentimental differences. Therefore, a multi-modal fusion attention sentiment analysis based on MSC to address this challenge is proposed. Firstly, the sentiment analysis fusion strategy based on multi-modal fusion is studied, which can fully utilise the information of multi-modal inputs such as text, audio, and video, thereby gaining a more comprehensive understanding and recognition of sentiments. Secondly, a sentiment analysis model based on multi-modal fusion attention is constructed, which focuses on the key information of multi-modal inputs to achieve an accurate recognition of mixed sentiments. The experimental results show that the proposed method outperforms existing sentiment analysis methods on both datasets, with F1 values of 83.17 and 84.19, accuracy of 39.15 and 39.98, and errors of 0.516 and 0.524, respectively. The accuracy range is 95.38%–99.89%, verifying the superiority of the method in sentiment analysis. It can be seen that this method provides a more effective and reliable MSC solution, which has practical significance for improving the accuracy and recall of sentiment analysis.

Fixed control parameters and various optimisation objectives significantly limit the robot control performance. To address such issues, a parameter adaptive controller based on deep reinforcement learning is introduced firstly to adjust control parameters according to the real-time system state. Further, multiple evaluation mechanisms are constructed to take account of optimisation objectives so that the controller can adapt to different control performance indexes by different evaluation mechanisms. Finally, the target pedestrian tracking control with mobile robots is selected as the validation case study, and the Proportional-Derivative Controller is chosen as the foundation controller. Several simulation and experimental examples are designed, and the results demonstrate that the proposed method shows satisfactory performance while taking account of multiple optimisation objectives.

An investigation on the effect of mental activity in quality perception is presented using simultaneous measurement of electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS), in a subject-independent approach. Building a subject-independent model is a harder problem due to noise and high EEG variability between individuals, correlated components analysis (CorrCA) have been proposed to extract significant correlated components for a single subject that experiences multiple identical trials; this is done by identifying spatio-temporal patterns of activity that are well preserved across trials. The aim is to build a model based on neurophysiological data to assess text-to-speech quality. In order to build a subject independent model, we extended the use of CorrCA such that it can be applied to the subject independent model. The authors used two preprocessing steps, namely the subject dependent and the stimulus dependent preprocessing. The second preprocessing used the denoising source separation (DSS) to remove noise/artefact that are subject specific. The discrete convolution is used for data fusion and the support vector machine for regression. With the proposed model, the fusion of EEG and fNIRS performs better than single modality. Using our defined regression accuracy metrics, the authors obtained accuracy of 81.346% for overall impression, 83.28% for valence and 89.714% for arousal. The model compete the baseline that is subject dependent.

Magnetic Resonance Imaging (MRI) based Autism Spectrum Disorder (ASD) detection approaches face various challenges due to variations in brain connectivity patterns, limited sample sizes, and heterogeneity of available data. These challenges make it hard to find consistent imaging markers. To address these issues, researchers have focused on advanced analysis methods, such as multi-modal imaging techniques and graph-based approaches to gain a comprehensive understanding of ASD neurobiology. However, existing graph-based approaches for ASD detection have primarily focused on pairwise similarities between individuals, neglecting individual characteristics and features. A novel framework to detect ASD using a Multi-Scale Enhanced Graph Convolutional Network (MSE-GCN). The framework combines the functional connectivity of resting-state functional MRI (rs-fMRI) with non-imaging phenotype data from Autism Brain Imaging Data Exchange-I (ABIDE-I). The framework uses MSE-GCN to represent individuals as node in a population graph. Each node corresponds to an individual and connects to feature vectors from imaging data. Edge weights between nodes are assigned to integrate phenotypic information. Then, the multiple parallel GCN layers are designed using random walk embedding. The output of these GCN layers is then combined in the fully connected layer to detect ASD effectively. The performance of the framework is evaluated using the ABIDE-I dataset. In addition, Recursive Feature Elimination and Multilayer Perceptron are utilised for feature selection. The outcome of this approach shows more than 10% advancement in accuracy, achieving an accuracy of 83% by incorporating phenotypic data in conjunction with MRI data within a GCN.