Cognitive Computation最新文献

In medical imaging, traditional methods have long been relied upon. However, the integration of Generative Adversarial Networks (GANs) has sparked a paradigm shift, ushering in a new era of innovation. Our comprehensive investigation explores the groundbreaking impact of GANs on medical imaging, examining the evolution from traditional techniques to GAN-driven approaches. Through meticulous analysis, we dissect various aspects of GANs, encompassing their taxonomy, historical progression, and diverse iterations such as Self-Attention GANs (SAGAN), Conditional GANs, and Progressive Growing GANs (PGGAN). Complemented by a practical case study, we scrutinize the extensive applications of GANs, spanning image generation, reconstruction, enhancement, segmentation, and super-resolution. Despite promising prospects, enduring challenges including data scarcity, interpretability issues, and ethical concerns persist. Looking ahead, we anticipate advancements in personalized and pathological image generation, cross-modal synthesis, real-time interactive image generation, and enhanced anomaly detection. Through this review, we underscore the transformative potential of GANs in reshaping medical imaging practices, while also outlining avenues for future research endeavors.

Current brain science reveals that the connectivity patterns of the human brain are constantly changing when performing different tasks. Thus, brain effective connectivity networks based on non-stationary assumption can describe such neurodynamics better than the ones based on stationary assumption. However, existing methods for inferring non-stationary brain effective connectivity networks are committed to estimating the change points and network structures simultaneously. It is even worse that these methods will inevitably focus on one part of the estimation process and lead to the deviation of the results obtained by the other part. Then, the construction results of non-stationary brain effective connectivity networks cannot accurately reflect the real brain dynamics. In this paper, a novel approach to constructing non-stationary brain effective connectivity networks is proposed, namely CPD-NSL. It involves two stages including change point detection and network structure learning. In the first stage, the latent block model is used, and then the improved forward-backward search method is used to construct the stationary networks between adjacent change points in the network structure learning part. Finally, the constructed stationary networks are arranged in chronological order to obtain the final time-varying brain effective connectivity network. CPD-NSL is validated using simulated data as well as real fMRI data from HCP public datasets. The results show that CPD-NSL can restore the real network more accurately and consume less time. Experimental results on both simulated and real data demonstrate the effectiveness of the proposed method in constructing non-stationary state brain effective connectivity networks.

As is well known that time delays are inevitable in practice due to the finite switching speed of amplifiers and information transmission between neurons. So the study on the Hopf bifurcation of delayed neural networks has aroused extensive attention in recent years. However, it’s worth mentioning that only the communication delays between neurons were generally considered in most existing relevant literatures. Actually, it has been proven that a kind of so-called leakage delays cannot be ignored because the self-decay process of a neuron’s action potential is not instantaneous in hardware implementation of neural networks. Though leakage delays have been taken into account in a few more recent works concerning the Hopf bifurcation of fractional-order bidirectional associative memory neural networks, the addressed neural networks were low-dimension or the involved time delays were single. In this paper, we propose a five-neuron fractional-order bidirectional associative memory neural network model, which includes leakage delays and discrete communication delays to meet the characteristics of real neural networks better. Then we use the stability theory of fractional differential equations and Hopf bifurcation theory to investigate its dynamic behavior of Hopf bifurcation. The Hopf bifurcation of the proposed model are studied by taking the involved two different leakage delays as the bifurcation parameter respectively, and two kinds of sufficient conditions for Hopf bifurcation are obtained. A numerical example as well as its simulation plots and phase portraits are given at last. Our results indicate that a Hopf bifurcation rises near the zero equilibrium point when the leakage delay reaches its critical value which is given by an explicit formula. Particularly, the results of numerical simulations show that the leakage delay would narrow the stability region of the proposed system and make the Hopf bifurcation occur earlier.

With the burgeoning global demand for seafood, potential solutions like aquaculture are increasingly significant, provided they address issues like pollution and food security challenges in a sustainable manner. However, significant obstacles such as disease outbreaks and inaccurate biomass estimation underscore the need for optimized solutions. This paper proposes “Fish-Sense”, a deep learning-based pipeline inspired by the human visual system’s ability to recognize and classify objects, developed in conjunction with fish farms, aiming to enhance disease detection and biomass estimation in the aquaculture industry. Our automated framework is two-pronged: one module for biomass estimation using deep learning algorithms to segment fish, classify species, and estimate biomass; and another for disease symptom detection symptoms, employing deep learning algorithms to classify fish into healthy and unhealthy categories, and subsequently identifying symptoms and locations of bacterial infections if a fish is classified as unhealthy. To overcome data scarcity in this field, we have created four novel real-world datasets for fish segmentation, health classification, species classification, and fish part segmentation. Our biomass estimation algorithms demonstrated substantial accuracy across five species, and the health classification. These algorithms provide a foundation for the development of industrial software solutions to improve fish health monitoring in aquaculture farms. Our integrated pipeline facilitates the transition from research to real-world applications, potentially encouraging responsible aquaculture practices. Nevertheless, these advancements must be seen as part of a comprehensive strategy aimed at improving the aquaculture industry’s sustainability and efficiency, in line with the United Nations’ Sustainable Development Goals’ evolving interpretations. The code, trained models, and the data for this project can be obtained from the following GitHub repository: https://github.com/Vision-At-SEECS/Fish-Sense.

Emotion analysis, an integral aspect of human–machine interactions, has witnessed significant advancements in recent years. With the rise of multimodal data sources such as speech, text, and images, there is a profound need for a comprehensive review of pivotal elements within this domain. Our paper delves deep into the realm of emotion analysis, examining multimodal data sources encompassing speech, text, images, and physiological signals. We provide a curated overview of relevant literature, academic forums, and competitions. Emphasis is laid on dissecting unimodal processing methods, including preprocessing, feature extraction, and tools across speech, text, images, and physiological signals. We further discuss the nuances of multimodal data fusion techniques, spotlighting early, late, model, and hybrid fusion strategies. Key findings indicate the essentiality of analyzing emotions across multiple modalities. Detailed discussions on emotion elicitation, expression, and representation models are presented. Moreover, we uncover challenges such as dataset creation, modality synchronization, model efficiency, limited data scenarios, cross-domain applicability, and the handling of missing modalities. Practical solutions and suggestions are provided to address these challenges. The realm of multimodal emotion analysis is vast, with numerous applications ranging from driver sentiment detection to medical evaluations. Our comprehensive review serves as a valuable resource for both scholars and industry professionals. It not only sheds light on the current state of research but also highlights potential directions for future innovations. The insights garnered from this paper are expected to pave the way for subsequent advancements in deep multimodal emotion analysis tailored for real-world deployments.

The precise segmentation of skin lesion in dermoscopic images is essential for the early detection of skin cancer. However, the irregular shapes of the lesions, the absence of sharp edges, the existence of artifacts like hair follicles, and marker color make this task difficult. Currently, fully connected networks (FCNs) and U-Nets are the most commonly used techniques for melanoma segmentation. However, as the depth of these neural network models increases, they become prone to various challenges. The most pertinent of these challenges are the vanishing gradient problem and the parameter redundancy problem. These can result in a decline in Jaccard index of the segmentation model. This study introduces a novel end-to-end trainable network designed for skin lesion segmentation. The proposed methodology consists of an encoder-decoder, a region-aware attention approach, and guided loss function. The trainable parameters are reduced using depth-wise separable convolution, and the attention features are refined using a guided loss, resulting in a high Jaccard index. We assessed the effectiveness of our proposed RA-Net on four frequently utilized benchmark datasets for skin lesion segmentation: ISIC 2016, ISIC 2017, ISIC 2018, and PH2. The empirical results validate that our method achieves state-of-the-art performance, as indicated by a notably high Jaccard index.

Ransomware is a significant security threat that poses a serious risk to the security of smartphones, and its impact on portable devices has been extensively discussed in a number of research papers. In recent times, this threat has witnessed a significant increase, causing substantial losses for both individuals and organizations. The emergence and widespread occurrence of diverse forms of ransomware present a significant impediment to the pursuit of reliable security measures that can effectively combat them. This constitutes a formidable challenge due to the dynamic nature of ransomware, which renders traditional security protocols inadequate, as they might have a high false alarm rate and exert significant processing demands on mobile devices that are restricted by limited battery life, CPU, and memory. This paper proposes a novel intelligent method for detecting ransomware that is based on a hybrid multi-solution binary JAYA algorithm with a single-solution simulated annealing (SA). The primary objective is to leverage the exploitation power of SA in supporting the exploration power of the binary JAYA algorithm. This approach results in a better balance between global and local search milestones. The empirical results of our research demonstrate the superiority of the proposed SMO-BJAYA-SA-SVM method over other algorithms based on the evaluation measures used. The proposed method achieved an accuracy rate of 98.7%, a precision of 98.6%, a recall of 98.7%, and an F1 score of 98.6%. Therefore, we believe that our approach is an effective method for detecting ransomware on portable devices. It has the potential to provide a more reliable and efficient solution to this growing security threat.

Differences in data size per class, also known as imbalanced data distribution, have become a common problem affecting data quality. Big Data scenarios pose a new challenge to traditional imbalanced classification algorithms, since they are not prepared to work with such amount of data. Split data strategies and lack of data in the minority class due to the use of MapReduce paradigm have posed new challenges for tackling the imbalance between classes in Big Data scenarios. Ensembles have been shown to be able to successfully address imbalanced data problems. Smart Data refers to data of enough quality to achieve high-performance models. The combination of ensembles and Smart Data, achieved through Big Data preprocessing, should be a great synergy. In this paper, we propose a novel Smart Data driven Decision Trees Ensemble methodology for addressing the imbalanced classification problem in Big Data domains, namely SD_DeTE methodology. This methodology is based on the learning of different decision trees using distributed quality data for the ensemble process. This quality data is achieved by fusing random discretization, principal components analysis, and clustering-based random oversampling for obtaining different Smart Data versions of the original data. Experiments carried out in 21 binary adapted datasets have shown that our methodology outperforms random forest.

Attention-deficit/hyperactivity disorder (ADHD) is a behavioral disorder in children that can persist into adulthood if not treated. Early diagnosis of this condition is crucial for effective treatment. The database includes 61 children with attention-deficit/hyperactivity disorder and 60 healthy children as a control group. To diagnose children with ADHD, features were first extracted from EEG signals. Next, a convolutional neural network model was trained, and a new residual network was introduced. The two proposed models were evaluated using tenfold cross-validation on the test data. The average accuracy and F1 score were 92.52% and 93.6%, respectively, for the convolutional model and 96.8% and 97.1% for the ResNet model on the epoch data, respectively. On the other hand, accuracy for subject-based prediction was 96.5% for the convolution model and 98.6% for the modified ResNet model. Accuracy, precision, recall, and F1 score for the proposed ResNet model are better than the convolution model proposed in previous studies and better than the proposed model in the literature. This work presents a paradigm shift in the cognitive-inspired domain by introducing a novel ResNet model for ADHD diagnosis. The model’s exceptional accuracy, exceeding conventional methods, showcases its potential as a biologically inspired tool. This opens avenues for exploring the neurological underpinnings of ADHD because the model can be used for the manifold learning of EEG signals. Analyzing the proposed network can lead to a deeper understanding of EEG, bridging the gap between artificial intelligence and cognitive neuroscience. The paper’s innovative approach has far-reaching implications, offering a concrete application of cognitive principles to improve mental health diagnostics in children. It is important to note that the data were augmented and the classification model is based on a single experiment containing a very small number of children but the results, and accuracy of classification, are based on classifying augmented data samples that compose the EEG signals of this small number of individuals. It is prudent to undertake a comprehensive investigation into the efficacy of these models across a broad cohort of subjects.