RESUMO
Heavy metal contamination of soils has been a global environmental issue over the past decades, threatening food security and human health. Understanding the migration and transformation of heavy metals in soils is critical for restoring an impaired environment and developing sustainable agriculture, particularly in the face of global warming. However, little effort has been devoted to investigating the impact of elevated temperatures on the migration and distribution of exogenous heavy metals in soils. This study experimented with a 180-day incubation at 15 °C, 30 °C, and 45 °C with an arable soil (Alfisol) of Huang-Huai-Hai River Basin, China, which was initially spiked with copper (Cu). A comparison of the results revealed that the percentage of soil water-soluble Cu doubled at 45 °C compared with 15 °C. The percentage of protein-like substances in dissolved organic matter (DOM) was the highest at 45 °C, suggesting that proteinaceous components play a more significant role in controlling the dissolution of Cu into DOM. Moreover, by sequential extraction and micro-X-ray fluorescence (µ-XRF), Cu was facilitatively transformed from exchangeable, and specifically adsorbed fractions, to iron (Fe)/manganese (Mn) oxides bound species by 7.75%23.63% with the elevation of temperature from 15 °C to 45 °C. The conversion of Cu speciation is attributed to the significant release of organic carbon from Fe/Mn oxides, especially the Mn oxide components, which are available for Cu binding. The findings of this work will provide an in-depth understanding of the fate of Cu in soils, which is fundamental for the risk assessment and remediation of Cu-polluted soils in the Huang-Huai-Hai River Basin under the context of global warming.
Assuntos
Metais Pesados , Poluentes do Solo , Humanos , Cobre/metabolismo , Solo/química , Temperatura , Metais Pesados/análise , Óxidos , Poluentes do Solo/análiseRESUMO
BACKGROUND: Frogeye leaf spot (FLS) is a destructive fungal disease that affects soybean production. The most economical and effective strategy to control FLS is the use of resistant cultivars. However, the use of a limited number of resistant loci in FLS management will be countered by the emergence of new high-virulence Cercospora sojina races. Therefore, we identified quantitative trait loci (QTL) that control resistance to FLS and identified novel resistant genes using a genome-wide association study (GWAS) on 234 Chinese soybean cultivars. RESULTS: A total of 30,890 single nucleotide polymorphism (SNP) markers were used to estimate linkage disequilibrium (LD) and population structure. The GWAS results showed four loci (p < 0.0001) distributed over chromosomes (Chr.) 5 and 20, that are significantly associated with FLS resistance. No previous studies have reported resistance loci in these regions. Subsequently, 45 genes in the two resistance-related haplotype blocks were annotated. Among them, Glyma20g31630 encoding pyruvate dehydrogenase (PDH), Glyma05g28980, which encodes mitogen-activated protein kinase 7 (MPK7), and Glyma20g31510, Glyma20g31520 encoding calcium-dependent protein kinase 4 (CDPK4) in the haplotype blocks deserves special attention. CONCLUSIONS: This study showed that GWAS can be employed as an effective strategy for identifying disease resistance traits in soybean and narrowing SNPs and candidate genes. The prediction of candidate genes in the haplotype blocks identified by disease resistance loci can provide a useful reference to study systemic disease resistance.
Assuntos
Cercospora/patogenicidade , Resistência à Doença/genética , Glycine max/genética , Doenças das Plantas/imunologia , Polimorfismo de Nucleotídeo Único/genética , Locos de Características Quantitativas/genética , Estudo de Associação Genômica Ampla , Genótipo , Haplótipos , Modelos Lineares , Desequilíbrio de Ligação , Fenótipo , Doenças das Plantas/microbiologia , Glycine max/imunologia , Glycine max/microbiologia , VirulênciaRESUMO
In sustainable smart cities, power saving is a severe challenge in the energy-constrained Internet of Things (IoT). Efficient utilization of limited multiple non-overlap channels and time resources is a promising solution to reduce the network interference and save energy consumption. In this paper, we propose a joint channel allocation and time slot optimization solution for IoT. First, we propose a channel ranking algorithm which enables each node to rank its available channels based on the channel properties. Then, we propose a distributed channel allocation algorithm so that each node can choose a proper channel based on the channel ranking and its own residual energy. Finally, the sleeping duration and spectrum sensing duration are jointly optimized to maximize the normalized throughput and satisfy energy consumption constraints simultaneously. Different from the former approaches, our proposed solution requires no central coordination or any global information that each node can operate based on its own local information in a total distributed manner. Also, theoretical analysis and extensive simulations have validated that when applying our solution in the network of IoT: (i) each node can be allocated to a proper channel based on the residual energy to balance the lifetime; (ii) the network can rapidly converge to a collision-free transmission through each node's learning ability in the process of the distributed channel allocation; and (iii) the network throughput is further improved via the dynamic time slot optimization.
RESUMO
Depression ranks among the most prevalent mood-related psychiatric disorders. Existing clinical diagnostic approaches relying on scale interviews are susceptible to individual and environmental variations. In contrast, the integration of neuroimaging techniques and computer science has provided compelling evidence for the quantitative assessment of major depressive disorder (MDD). However, one of the major challenges in computer-aided diagnosis of MDD is to automatically and effectively mine the complementary cross-modal information from limited datasets. In this study, we proposed a few-shot learning framework that integrates multi-modal MRI data based on contrastive learning. In the upstream task, it is designed to extract knowledge from heterogeneous data. Subsequently, the downstream task is dedicated to transferring the acquired knowledge to the target dataset, where a hierarchical fusion paradigm is also designed to integrate features across inter- and intra-modalities. Lastly, the proposed model was evaluated on a set of multi-modal clinical data, achieving average scores of 73.52% and 73.09% for accuracy and AUC, respectively. Our findings also reveal that the brain regions within the default mode network and cerebellum play a crucial role in the diagnosis, which provides further direction in exploring reproducible biomarkers for MDD diagnosis.
Assuntos
Transtorno Depressivo Maior , Humanos , Transtorno Depressivo Maior/diagnóstico por imagem , Aprendizagem , Imageamento por Ressonância Magnética , Neuroimagem , AfetoRESUMO
AIMS: To extract vertex-wise features of the hippocampus and amygdala in Parkinson's disease (PD) with mild cognitive impairment (MCI) and normal cognition (NC) and further evaluate their discriminatory efficacy. METHODS: High-resolution 3D-T1 data were collected from 68 PD-MCI, 211 PD-NC, and 100 matched healthy controls (HC). Surface geometric features were captured using surface conformal representation, and surfaces were registered to a common template using fluid registration. The statistical tests were performed to detect differences between groups. The disease-discriminatory ability of features was also tested in the ensemble classifiers. RESULTS: The amygdala, not the hippocampus, showed significant overall differences among the groups. Compared with PD-NC, the right amygdala in MCI patients showed expansion (anterior cortical, anterior amygdaloid, and accessory basal areas) and atrophy (basolateral ventromedial area) subregions. There was notable atrophy in the right CA1 and hippocampal subiculum of PD-MCI. The accuracy of classifiers with multivariate morphometry statistics as features exceeded 85%. CONCLUSION: PD-MCI is associated with multiscale morphological changes in the amygdala, as well as subtle atrophy in the hippocampus. These novel metrics demonstrated the potential to serve as biomarkers for PD-MCI diagnosis. Overall, these findings from this study help understand the role of subcortical structures in the neuropathological mechanisms of PD cognitive impairment.
Assuntos
Disfunção Cognitiva , Doença de Parkinson , Humanos , Doença de Parkinson/patologia , Imageamento por Ressonância Magnética , Disfunção Cognitiva/patologia , Cognição , Hipocampo/diagnóstico por imagem , Hipocampo/patologia , Atrofia/complicações , Atrofia/patologiaRESUMO
Increasing epidemiological evidence has shown that PM2.5 exposure is significantly associated with the occurrence of osteoporosis. It has been well demonstrated that PM2.5 exposure enhanced the differentiation and function of osteoclasts by indirectly causing chronic inflammation, while the mechanism in osteoblasts remains unclear. In our study, toxic effects were evaluated by direct exposure of 20-80⯵g/ml PM2.5 to MC3T3-E1 cells and BMSCs. The results showed that PM2.5 exposure did not affect cell viability via proliferation and apoptosis, but significantly inhibited osteoblast differentiation in a dose-dependent manner. Osteogenic transcription factors Runx2 and Sp7 and other biomarkers Alp and Ocn decreased after PM2.5 exposure. RNA-seq revealed TGF-ß signaling was involved in PM2.5 exposure inhibited osteoblast differentiation, which led to P-Smad1/5 and P-Smad2 reduction in the nucleus by increasing the ubiquitination and degradation of Smad4. At last, the inflammation response increased in MC3T3-E1 cells with PM2.5 exposure. Moreover, the mRNA levels of Mmp9 increased in bone marrow-derived macrophage cells treated with the conditional medium collected from MC3T3-E1 cells exposed to PM2.5. Overall, these results indicated that PM2.5 exposure inhibits osteoblast differentiation and concurrently increases the maturation of osteoclasts. Our study provides in-depth mechanistic insights into the direct impact of PM2.5 exposure on osteoblast, which would indicate the unrecognized role of PM2.5 on osteoporosis.
Assuntos
Diferenciação Celular , Osteoblastos , Material Particulado , Proteína Smad4 , Ubiquitinação , Osteoblastos/efeitos dos fármacos , Osteoblastos/metabolismo , Animais , Diferenciação Celular/efeitos dos fármacos , Proteína Smad4/metabolismo , Proteína Smad4/genética , Camundongos , Material Particulado/toxicidade , Ubiquitinação/efeitos dos fármacos , Transdução de Sinais/efeitos dos fármacos , Osteogênese/efeitos dos fármacos , Osteoclastos/efeitos dos fármacos , Osteoclastos/metabolismo , Poluentes Atmosféricos/toxicidade , Linhagem Celular , Sobrevivência Celular/efeitos dos fármacos , Fator de Crescimento Transformador beta/metabolismo , Subunidade alfa 1 de Fator de Ligação ao Core/metabolismo , Subunidade alfa 1 de Fator de Ligação ao Core/genética , Metaloproteinase 9 da Matriz/metabolismo , Metaloproteinase 9 da Matriz/genética , Proteólise/efeitos dos fármacosRESUMO
Facial expressions have been widely used for depression recognition because it is intuitive and convenient to access. Pupil diameter contains rich emotional information that is already reflected in facial video streams. However, the spatiotemporal correlation between pupillary changes and facial behavior changes induced by emotional stimuli has not been explored in existing studies. This paper presents a novel multimodal fusion algorithm - Trial Selection Tensor Canonical Correlation Analysis (TSTCCA) to optimize the feature space and build a more robust depression recognition model, which innovatively combines the spatiotemporal relevance and complementarity between facial expression and pupil diameter features. TSTCCA explores the interaction between trials and obtains an effective fusion representation of two modalities from a trial subset related to depression. The experimental results show that TSTCCA achieves the highest accuracy of 78.81% with the subset of 25 trials.
RESUMO
Effective estimation of brain network connectivity enables better unraveling of the extraordinary complexity interactions of brain regions and helps in auxiliary diagnosis of psychiatric disorders. Considering different modalities can provide comprehensive characterizations of brain connectivity, we propose the message-passing-based nonlinear network fusion (MP-NNF) algorithm to estimate multimodal brain network connectivity. In the proposed method, the initial functional and structural networks were computed from fMRI and DTI separately. Then, we update every unimodal network iteratively, making it more similar to the others in every iteration, and finally converge to one unified network. The estimated brain connectivities integrate complementary information from multiple modalities while preserving their original structure, by adding the strong connectivities present in unimodal brain networks and eliminating the weak connectivities. The effectiveness of the method was evaluated by applying the learned brain connectivity for the classification of major depressive disorder (MDD). Specifically, 82.18% classification accuracy was achieved even with the simple feature selection and classification pipeline, which significantly outperforms the competing methods. Exploration of brain connectivity contributed to MDD identification suggests that the proposed method not only improves the classification performance but also was sensitive to critical disease-related neuroimaging biomarkers.
RESUMO
Recently, psychophysiological computing has received considerable attention. Due to easy acquisition at a distance and less conscious initiation, gait-based emotion recognition is considered as a valuable research branch in the field of psychophysiological computing. However, most existing methods rarely explore the spatio-temporal context of gait, which limits the ability to capture the higher-order relationship between emotion and gait. In this paper, we utilize a range of research, including psychophysiological computing and artificial intelligence, to propose an integrated emotion perception framework called EPIC, which can find novel joint topology and generate thousands of synthetic gaits by spatio-temporal interaction context. First, we analyze the joint coupling among non-adjacent joints by calculating Phase Lag Index (PLI), which can discover the latent connection among body joints. Second, to synthesize more sophisticated and accurate gait sequences, we explore the effect of spatio-temporal constraints, and propose a new loss function that utilizes the Dynamic Time Warping (DTW) algorithm and pseudo-velocity curve to constrain the output of Gated Recurrent Units (GRU). Finally, Spatial Temporal Graph Convolution Networks (ST-GCN) is used to classify emotions using the generation and the real data. Experimental results demonstrate our approach achieves the accuracy of 89.66%, and outperforms the state-of-the-art methods on Emotion-Gait dataset.
RESUMO
Depression is one of the most common mental disorders, with sleep disturbances as typical symptoms. With the popularity of wearable devices increasing in recent years, more and more people wear portable devices to track sleep quality. Based on this, we believe that depression detection through wearable sleep data is more intelligent and economical. However, the majority of wearable devices face the problem of missing data during the data collection process. Otherwise, most existing studies of depression identification focus on the utilization of complex data, making it difficult to generalize and susceptible to noise interference. To address these issues, we propose a systematic ensemble classification model for depression (ECD). For the missing data problem of wearable devices, we design an improved GAIN method to further control the generation range of interpolated values, which can achieve a more reasonable treatment of missing values. Compared with the original GAIN approach, the improved method shows a 28.56% improvement when using MAE as the metric. For depression recognition, we use ensemble learning to construct a depression classification model which combines five classification models, including SVM, KNN, LR, CBR, and DT. Ensemble learning can improve the model's robustness and generalization. The voting mechanism is used in several places to improve noise immunity. The final classification model performed great on the dataset, with a precision of 92.55% and a recall of 91.89%. These results illustrate how efficient this method is in automatically detecting depression.
RESUMO
Introduction: Deep-learn methods based on convolutional neural networks (CNNs) have demonstrated impressive performance in depression analysis. Nevertheless, some critical challenges need to be resolved in these methods: (1) It is still difficult for CNNs to learn long-range inductive biases in the low-level feature extraction of different facial regions because of the spatial locality. (2) It is difficult for a model with only a single attention head to concentrate on various parts of the face simultaneously, leading to less sensitivity to other important facial regions associated with depression. In the case of facial depression recognition, many of the clues come from a few areas of the face simultaneously, e.g., the mouth and eyes. Methods: To address these issues, we present an end-to-end integrated framework called Hybrid Multi-head Cross Attention Network (HMHN), which includes two stages. The first stage consists of the Grid-Wise Attention block (GWA) and Deep Feature Fusion block (DFF) for the low-level visual depression feature learning. In the second stage, we obtain the global representation by encoding high-order interactions among local features with Multi-head Cross Attention block (MAB) and Attention Fusion block (AFB). Results: We experimented on AVEC2013 and AVEC2014 depression datasets. The results of AVEC 2013 (RMSE = 7.38, MAE = 6.05) and AVEC 2014 (RMSE = 7.60, MAE = 6.01) demonstrated the efficacy of our method and outperformed most of the state-of-the-art video-based depression recognition approaches. Discussion: We proposed a deep learning hybrid model for depression recognition by capturing the higher-order interactions between the depression features of multiple facial regions, which can effectively reduce the error in depression recognition and gives great potential for clinical experiments.
RESUMO
Shading in combination with extended photoperiods can cause exaggerated stem elongation (ESE) in soybean, leading to lodging and reduced yields when planted at high-density in high-latitude regions. However, the genetic basis of plant height in adaptation to these regions remains unclear. Here, through a genome-wide association study, we identify a plant height regulating gene on chromosome 13 (PH13) encoding a WD40 protein with three main haplotypes in natural populations. We find that an insertion of a Ty1/Copia-like retrotransposon in the haplotype 3 leads to a truncated PH13H3 with reduced interaction with GmCOP1s, resulting in accumulation of STF1/2, and reduced plant height. In addition, PH13H3 allele has been strongly selected for genetic improvement at high latitudes. Deletion of both PH13 and its paralogue PHP can prevent shade-induced ESE and allow high-density planting. This study provides insights into the mechanism of shade-resistance and offers potential solutions for breeding high-yielding soybean cultivar for high-latitude regions.
Assuntos
Estudo de Associação Genômica Ampla , Glycine max , Glycine max/genética , Melhoramento Vegetal , Fenótipo , RetroelementosRESUMO
In recent years, the incidence of depression is rising rapidly worldwide, but large-scale depression screening is still challenging. Gait analysis provides a non-contact, low-cost, and efficient early screening method for depression. However, the early screening of depression based on gait analysis lacks sufficient effective sample data. In this paper, we propose a skeleton data augmentation method for assessing the risk of depression. First, we propose five techniques to augment skeleton data and apply them to depression and emotion datasets. Then, we divide augmentation methods into two types (non-noise augmentation and noise augmentation) based on the mutual information and the classification accuracy. Finally, we explore which augmentation strategies can capture the characteristics of human skeleton data more effectively. Experimental results show that the augmented training dataset that retains more of the raw skeleton data properties determines the performance of the detection model. Specifically, rotation augmentation and channel mask augmentation make the depression detection accuracy reach 92.15% and 91.34%, respectively.
Assuntos
Depressão , Esqueleto , Depressão/diagnóstico , Marcha , HumanosRESUMO
Dissolved organic matter (DOM) is one of the most active soil components, which plays pivotal roles in the migration and fate of heavy metals in soils. The interactions of heavy metals with DOM are controlled by the structure and properties of DOM. The changes of temperature have a significant effect on the content and composition of DOM and thus may affect the binding nature of heavy metals with DOM. In the current study, we conducted a 180-d incubation experiment with an arable soil at temperatures of 15, 30 and 45 °C. Fluorescence spectroscopy was used to examine the composition of DOM and two-dimensional correlation spectroscopy was applied to determine the binding intensity and sequence between cadmium (Cd) with DOM. Two humic-like substances (C1, C3) and a protein-like substance (C2) were identified from soil DOM. Elevated temperature changed the characteristic and structure of DOM. The humification degree and aromaticity of DOM increased from 15 °C to 30 °C but decreased at high temperature (45 °C). The alterations in temperature exert no impact on the type of organic functional groups in DOM binding with Cd. However, elevated temperature changed the binding sequence of Cd with DOM fractions. Polysaccharide, phenolic, and aromatic groups exhibited the fastest response to Cd at 15, 30, and 45 °C, respectively. These observations would provide a better understanding on the environmental behavior of Cd in arable soils under the context of global warming.
Assuntos
Cádmio , Metais Pesados , Matéria Orgânica Dissolvida , Solo , TemperaturaRESUMO
Currently, depression has become a common mental disorder, especially among postgraduates. It is reported that postgraduates have a higher risk of depression than the general public, and they are more sensitive to contact with others. Thus, a non-contact and effective method for detecting people at risk of depression becomes an urgent demand. In order to make the recognition of depression more reliable and convenient, we propose a multi-modal gait analysis-based depression detection method that combines skeleton modality and silhouette modality. Firstly, we propose a skeleton feature set to describe depression and train a Long Short-Term Memory (LSTM) model to conduct sequence strategy. Secondly, we generate Gait Energy Image (GEI) as silhouette features from RGB videos, and design two Convolutional Neural Network (CNN) models with a new loss function to extract silhouette features from front and side perspectives. Then, we construct a multi-modal fusion model consisting of fusing silhouettes from the front and side views at the feature level and the classification results of different modalities at the decision level. The proposed multi-modal model achieved accuracy at 85.45% in the dataset consisting of 200 postgraduate students (including 86 depressive ones), 5.17% higher than the best single-mode model. The multi-modal method also shows improved generalization by reducing the gender differences. Furthermore, we design a vivid 3D visualization of the gait skeletons, and our results imply that gait is a potent biometric for depression detection.
Assuntos
Depressão , Análise da Marcha , Biometria , Depressão/diagnóstico , Marcha , Humanos , Redes Neurais de ComputaçãoRESUMO
Person re-identification (Re-ID) via gait features within 3D skeleton sequences is a newly-emerging topic with several advantages. Existing solutions either rely on hand-crafted descriptors or supervised gait representation learning. This paper proposes a self-supervised gait encoding approach that can leverage unlabeled skeleton data to learn gait representations for person Re-ID. Specifically, we first create self-supervision by learning to reconstruct unlabeled skeleton sequences reversely, which involves richer high-level semantics to obtain better gait representations. Other pretext tasks are also explored to further improve self-supervised learning. Second, inspired by the fact that motion's continuity endows adjacent skeletons in one skeleton sequence and temporally consecutive skeleton sequences with higher correlations (referred as locality in 3D skeleton data), we propose a locality-aware attention mechanism and a locality-aware contrastive learning scheme, which aim to preserve locality-awareness on intra-sequence level and inter-sequence level respectively during self-supervised learning. Last, with context vectors learned by our locality-aware attention mechanism and contrastive learning scheme, a novel feature named Constrastive Attention-based Gait Encodings (CAGEs) is designed to represent gait effectively. Empirical evaluations show that our approach significantly outperforms skeleton-based counterparts by 15-40 percent Rank-1 accuracy, and it even achieves superior performance to numerous multi-modal methods with extra RGB or depth information. Our codes are available at https://github.com/Kali-Hac/Locality-Awareness-SGE.
Assuntos
Algoritmos , Marcha , Humanos , EsqueletoRESUMO
Visual relationship detection (VRD) is one newly developed computer vision task, aiming to recognize relations or interactions between objects in an image. It is a further learning task after object recognition, and is important for fully understanding images even the visual world. It has numerous applications, such as image retrieval, machine vision in robotics, visual question answer (VQA), and visual reasoning. However, this problem is difficult since relationships are not definite, and the number of possible relations is much larger than objects. So the complete annotation for visual relationships is much more difficult, making this task hard to learn. Many approaches have been proposed to tackle this problem especially with the development of deep neural networks in recent years. In this survey, we first introduce the background of visual relations. Then, we present categorization and frameworks of deep learning models for visual relationship detection. The high-level applications, benchmark datasets, as well as empirical analysis are also introduced for comprehensive understanding of this task.
Assuntos
Redes Neurais de Computação , Percepção VisualRESUMO
Dissolved organic matter (DOM) represents the most mobile and reactive pool of soil organic matter (SOM). Climate changes, such as global warming and altered precipitation exert considerable influences on the quality and quantity of soil DOM. However, rare reports have focused on the interactive effects of soil warming and increased precipitation. In the present study, we conducted a 90-day incubation experiment to investigate how the concentration, source and chemical composition of DOM from an Alfisol respond to the variations of temperatures (15, 30 and 45 °C) and moistures (40%, 60%, and 80% of saturated soil water content). Four DOM components were identified through fluorescence excitation emission matrix (EEM)-parallel factor analysis (PARAFAC). Increased temperature alone aggravated the decomposition of plant-derived aromatic components (C2 and C4) but promoted the accumulation of microbial-derived aliphatic carbon (C1) and tryptophan-like component (C3). Increased fungi/bacteria ratio with warming was responsible for the decomposition of plant-derived components. Warming-induced disassociation of Ca-bearing mineral to colloidal Ca facilitated the accrual of microbial-derived aliphatic DOM. Humidification alone and humidification + warming significantly increased the concentration of DOM and the percentage of plant-derived aromatic carbon (C2, C4), which was attributed to the release of Fe-bearing mineral-OC. Based on the above findings along with the results of two-way ANOVA and Variation partition analysis, we infer that moisture will play a dominant role in regulating the chemical composition of DOM in Alfisols under both warming and humidification which in turn impact global C cycling and the ultimate climate.
Assuntos
Solo , Qualidade da Água , Carbono , Substâncias Húmicas/análise , Espectrometria de Fluorescência , Temperatura , ÁguaRESUMO
According to the WHO, the number of mental disorder patients, especially depression patients, has overgrown and become a leading contributor to the global burden of disease. With the rising of tools such as artificial intelligence, using physiological data to explore new possible physiological indicators of mental disorder and creating new applications for mental disorder diagnosis has become a new research hot topic. We present a multi-modal open dataset for mental-disorder analysis. The dataset includes EEG and recordings of spoken language data from clinically depressed patients and matching normal controls, who were carefully diagnosed and selected by professional psychiatrists in hospitals. The EEG dataset includes data collected using a traditional 128-electrodes mounted elastic cap and a wearable 3-electrode EEG collector for pervasive computing applications. The 128-electrodes EEG signals of 53 participants were recorded as both in resting state and while doing the Dot probe tasks; the 3-electrode EEG signals of 55 participants were recorded in resting-state; the audio data of 52 participants were recorded during interviewing, reading, and picture description.
Assuntos
Transtornos Mentais , Inteligência Artificial , Eletroencefalografia , Humanos , Transtornos Mentais/diagnóstico , Transtornos Mentais/fisiopatologiaRESUMO
Facial expression recognition has become a newly-emerging topic in recent decades, which has important value in the field of human-computer interaction. In this paper, we present a deep learning based approach, named frequency neural network (FreNet), for facial expression recognition. Different from convolutional neural network in spatial domain, FreNet inherits the advantages of processing image in frequency domain, such as efficient computation and spatial redundancy elimination. First, we propose the learnable multiplication kernel and construct multiple multiplication layers to learn features in frequency domain. Second, a summarization layer is proposed following multiplication layers to further yield high-level features. Third, based on the property of discrete cosine transform (DCT), we utilize multiplication layers and summarization layer to construct the Basic-FreNet, which can yield high-level features on the widely used DCT feature. Finally, to further achieve better performance on Basic-FreNet, we propose the Block-FreNet in which the weight-shared multiplication kernel is designed for feature learning and the block sub-sampling is designed for dimension reduction. The experimental results show that the Block-FreNet not only achieves superior performance, but also greatly reduces the computational cost. To our best knowledge, the proposed approach is the first attempt to fill in the blank of frequency based deep learning model for facial expression recognition.