RESUMO
Neuropathic pain is a significant and persistent issue for individuals with spinal cord injuries (SCI), severely impacting their quality of life. While changes at the peripheral and spinal levels are known to contribute to SCI-related pain, whether and how supraspinal centers contribute to post SCI chronic neuropathic pain is poorly understood. Here, we first validated delayed development of chronic neuropathic pain in mice with moderate contusion SCI. To identify supraspinal regions involved in the pathology of neuropathic pain after SCI, we next performed an activity dependent genetic screening and identified multiple cortical and subcortical regions that were activated by innocuous tactile stimuli at a late stage following contusion SCI. Notably, chemogenetic inactivation of pain trapped neurons in the lateral thalamus alleviated neuropathic pain and reduced tactile stimuli evoked cortical overactivation. Retrograde tracing showed that contusion SCI led to enhanced corticothalamic axonal sprouting and over-activation of corticospinal neurons. Mechanistically, ablation or silencing of corticospinal neurons prevented the establishment or maintenance of chronic neuropathic pain following contusion SCI. These results highlighted a corticospinal-lateral thalamic feed-forward loop whose activation is required for the development and maintenance of chronic neuropathic pain after SCI. Our data thus shed lights into the central mechanisms underlying chronic neuropathic pain associated with SCI and the development of novel therapeutic avenues to treat refractory pain caused by traumatic brain or spinal cord injuries.
RESUMO
The cerebral cortex performs computations via numerous six-layer modules. The operational dynamics of these modules were studied primarily in early sensory cortices using bottom-up computation for response selectivity as a model, which has been recently revolutionized by genetic approaches in mice. However, cognitive processes such as recall and imagery require top-down generative computation. The question of whether the layered module operates similarly in top-down generative processing as in bottom-up sensory processing has become testable by advances in the layer identification of recorded neurons in behaving monkeys. This review examines recent advances in laminar signaling in these two computations, using predictive coding computation as a common reference, and shows that each of these computations recruits distinct laminar circuits, particularly in layer 5, depending on the cognitive demands. These findings highlight many open questions, including how different interareal feedback pathways, originating from and terminating at different layers, convey distinct functional signals.
Assuntos
Córtex Cerebral , Cognição , Animais , Cognição/fisiologia , Córtex Cerebral/fisiologia , Humanos , Neurônios/fisiologia , Modelos Neurológicos , Vias Neurais/fisiologia , Rede Nervosa/fisiologia , Transdução de Sinais/fisiologiaRESUMO
Friction is the dominant factor restricting tracking accuracy and machining surface quality in mechanical systems such as machine tool feed-drive. Hence, friction modeling and compensation is an important method in accurate tracking control of CNC machine tools used for welding, 3D printing, and milling, etc. Many static and dynamic friction models have been proposed to compensate for frictional effects to reduce the tracking error in the desired trajectory and to improve the surface quality. However, most of them focus on the friction characteristics of the pre-sliding zone and low-speed sliding regions. These models do not fully describe friction in the case of insufficient lubrication or high acceleration and deceleration in machine tool systems. This paper presents a new nonlinear friction model that includes the typical Coulomb-Viscous friction, a nonlinear periodic harmonic friction term for describing the lead screw property in insufficient lubrication, and a functional component of acceleration for describing the friction lag caused by the acceleration and deceleration of the system. Experiments were conducted to compare the friction compensation performance between the proposed and the conventional friction models. Experimental results indicate that the root mean square and maximum absolute tracking error can be significantly reduced after applying the proposed friction model.
RESUMO
The bilateral-to-radial symmetry transition occurring during the development of the Arabidopsis thaliana female reproductive organ (gynoecium) is a crucial biological process linked to plant fertilization and seed production. Despite its significance, the cellular mechanisms governing the establishment and breaking of radial symmetry at the gynoecium apex (style) remain unknown. To fill this gap, we employed quantitative confocal imaging coupled with MorphoGraphX analysis, in vivo and in vitro transcriptional experiments, and genetic analysis encompassing mutants in two bHLH transcription factors necessary and sufficient to promote transition to radial symmetry, SPATULA (SPT) and INDEHISCENT (IND). Here, we show that defects in style morphogenesis correlate with defects in cell-division orientation and rate. We showed that the SPT-mediated accumulation of auxin in the medial-apical cells undergoing symmetry transition is required to maintain cell-division-oriented perpendicular to the direction of organ growth (anticlinal, transversal cell division). In addition, SPT and IND promote the expression of specific core cell-cycle regulators, CYCLIN-D1;1 (CYC-D1;1) and CYC-D3;3, to support progression through the G1 phase of the cell cycle. This transcriptional regulation is repressed by auxin, thus forming an incoherent feed-forward loop mechanism. We propose that this mechanism fine-tunes cell division rate and orientation with the morphogenic signal provided by auxin, during patterning of radial symmetry at the style.
RESUMO
PURPOSE: To investigate the effects of concurrent sensorimotor training (SMT) and transcranial direct current stimulation (tDCS) on the anticipatory and compensatory postural adjustments (APAs and CPAs) in patients with chronic low back pain (CLBP). METHOD: The interventions included (1) SMT plus tDCS and (2) SMT plus sham tDCS. Outcome measures were the normalized integrals of electromyography activity (NIEMG) during the phases of anticipatory and compensatory, and muscle onset latency. The investigated muscles were ipsilateral and contralateral multifidus (MF), transversus abdominus/internal oblique (TrA/IO), and gluteus medius (GM). RESULTS: Between-group comparisons demonstrated that ipsilateral TrA/IO NIEMG during CPA1 (p = 0.010) and ipsilateral GM NIEMG during CPA1 (p = 0.002) and CPA2 (p = 0.025) were significantly lower in the SMT combined with tDCS than in the control group. Furthermore, this group had greater NIEMG for contralateral GM during APA1 than the control group (p = 0.032). Moreover, the onset latency of contralateral TrA/IO was significantly earlier after SMT combined with tDCS (p = 0.011). CONCLUSIONS: Both groups that received SMT showed positive effects, but anodal tDCS had an added value over sham stimulation for improving postural control strategies in patients with CLBP. Indeed, SMT combined with tDCS leads to stronger APA and less demand for CPA. RCT REGISTRATION NUMBER: IRCT20220228054149N1. REGISTRATION DATE: 2022-04-04.
Evidence suggests that reduced excitability in the sensory and motor cortex is linked to chronic and recurring lower back pain.Increasing the excitability of these two areas using anodal transcranial direct current stimulation (tDCS), in conjunction with sensorimotor training (SMT), may improve anticipatory and compensatory postural control strategies.This study showed that the combination of SMT with tDCS targeting the sensory and motor cortex notably enhances motor preparation and refines postural control strategies in patients with chronic unilateral lumbar radiculopathy.Rehabilitation professionals are encouraged to integrate SMT with tDCS into treatment protocols to enhance the ability of individuals with back pain to handle postural disturbances in daily life, thereby potentially alleviating the persistence of their symptoms.Incorporating brain stimulation enhances the effectiveness of SMT for patients with chronic unilateral lumbar radiculopathy.
RESUMO
PURPOSE: The purpose of this study is to compare the reconstructed corneal power (RCP) by working backwards from the post-implantation spectacle refraction and toric intraocular lens power and to develop the models for mapping preoperative keratometry and total corneal power to RCP. METHODS: Retrospective single-centre study involving 442 eyes treated with a monofocal and trifocal toric IOL (Zeiss TORBI and LISA). Keratometry and total corneal power were measured preoperatively and postoperatively using IOLMaster 700. Feedforward neural network and multilinear regression models were derived to map keratometry and total corneal power vector components (equivalent power EQ and astigmatism components C0 and C45) to the respective RCP components. RESULTS: Mean preoperative/postoperative C0 for keratometry and total corneal power was -0.14/-0.08 dioptres and -0.30/-0.24 dioptres. All mean C45 components ranged between -0.11 and -0.20 dioptres. With crossvalidation, the neural network and regression models showed comparable results on the test data with a mean squared prediction error of 0.20/0.18 and 0.22/0.22 dioptres2 and on the training data the neural network models outperformed the regression models with 0.11/0.12 and 0.22/0.22 dioptres2 for predicting RCP from preoperative keratometry/total corneal power. CONCLUSIONS: Based on our dataset, both the feedforward neural network and multilinear regression models showed good precision in predicting the power vector components of RCP from preoperative keratometry or total corneal power. With a similar performance in crossvalidation and a simple implementation in consumer software, we recommend implementation of regression models in clinical practice.
RESUMO
To investigate the positive feed-forward regulatory mechanism of nitrate uptake by rice, its responses to various light and carbohydrates were compared. In order to measure nitrate uptake in real time, the non-invasive method was used. The results showed that net nitrate uptake increased in the light and decreased in the dark, and finally reached a steady state after about 5 h. Based on it, carbohydrates effects could be investigated without considering light effects. After sucrose addition for 2 h, net nitrate uptake increased by about 80% without a lag, while glucose, fructose and raffinose had a slight effect with a lag and other sugars had no effect. It provided an evidence that sucrose was a positive feed-forward signal molecule of nitrate uptake by rice roots. To further analyze the effect of sucrose on the expression of high affinity nitrate transporter genes OsNRT2.1, OsNRT2.2, OsNRT2.3a and OsNRT2.3b, qRT-PCR was used to further verify after treated with 10 mM sucrose. The results revealed that these genes expression was immediately up-regulated, which indicated that these genes were post transcriptionally regulated. Further, 15N exchange dynamics analyzed N transport. It is benefit for increasing nitrate uptake by rice and improving its yield.
Assuntos
Regulação da Expressão Gênica de Plantas , Nitratos , Oryza , Raízes de Plantas , Sacarose , Oryza/metabolismo , Oryza/genética , Raízes de Plantas/metabolismo , Raízes de Plantas/genética , Nitratos/metabolismo , Sacarose/metabolismo , Transporte Biológico , Proteínas de Plantas/metabolismo , Proteínas de Plantas/genética , Proteínas de Transporte de Ânions/metabolismo , Proteínas de Transporte de Ânions/genética , Luz , Transportadores de NitratoRESUMO
Previous studies have revealed that auditory processing is modulated during the planning phase immediately prior to speech onset. To date, the functional relevance of this pre-speech auditory modulation (PSAM) remains unknown. Here, we investigated whether PSAM reflects neuronal processes that are associated with preparing auditory cortex for optimized feedback monitoring as reflected in online speech corrections. Combining electroencephalographic PSAM data from a previous data set with new acoustic measures of the same participants' speech, we asked whether individual speakers' extent of PSAM is correlated with the implementation of within-vowel articulatory adjustments during /b/-vowel-/d/ word productions. Online articulatory adjustments were quantified as the extent of change in inter-trial formant variability from vowel onset to vowel midpoint (a phenomenon known as centering). This approach allowed us to also consider inter-trial variability in formant production and its possible relation to PSAM at vowel onset and midpoint separately. Results showed that inter-trial formant variability was significantly smaller at vowel midpoint than at vowel onset. PSAM was not significantly correlated with this amount of change in variability as an index of within-vowel adjustments. Surprisingly, PSAM was negatively correlated with inter-trial formant variability not only in the middle but also at the very onset of the vowels. Thus, speakers with more PSAM produced formants that were already less variable at vowel onset. Findings suggest that PSAM may reflect processes that influence speech acoustics as early as vowel onset and, thus, that are directly involved in motor command preparation (feedforward control) rather than output monitoring (feedback control).
RESUMO
Prostate Cancer is one of the most frequently occurring cancers in men, with a low survival rate if not early diagnosed. PI-RADS reading has a high false positive rate, thus increasing the diagnostic incurred costs and patient discomfort. Deep learning (DL) models achieve a high segmentation performance, although require a large model size and complexity. Also, DL models lack of feature interpretability and are perceived as "black-boxes" in the medical field. PCa-RadHop pipeline is proposed in this work, aiming to provide a more transparent feature extraction process using a linear model. It adopts the recently introduced Green Learning (GL) paradigm, which offers a small model size and low complexity. PCa-RadHop consists of two stages: Stage-1 extracts data-driven radiomics features from the bi-parametric Magnetic Resonance Imaging (bp-MRI) input and predicts an initial heatmap. To reduce the false positive rate, a subsequent stage-2 is introduced to refine the predictions by including more contextual information and radiomics features from each already detected Region of Interest (ROI). Experiments on the largest publicly available dataset, PI-CAI, show a competitive performance standing of the proposed method among other deep DL models, achieving an area under the curve (AUC) of 0.807 among a cohort of 1,000 patients. Moreover, PCa-RadHop maintains orders of magnitude smaller model size and complexity.
RESUMO
Our brain excels at recognizing objects, even when they flash by in a rapid sequence. However, the neural processes determining whether a target image in a rapid sequence can be recognized or not remains elusive. We used electroencephalography (EEG) to investigate the temporal dynamics of brain processes that shape perceptual outcomes in these challenging viewing conditions. Using naturalistic images and advanced multivariate pattern analysis (MVPA) techniques, we probed the brain dynamics governing conscious object recognition. Our results show that although initially similar, the processes for when an object can or cannot be recognized diverge around 180 ms post-appearance, coinciding with feedback neural processes. Decoding analyses indicate that gist perception (partial conscious perception) can occur at â¼120 ms through feedforward mechanisms. In contrast, object identification (full conscious perception of the image) is resolved at â¼190 ms after target onset, suggesting involvement of recurrent processing. These findings underscore the importance of recurrent neural connections in object recognition and awareness in rapid visual presentations.
Assuntos
Estado de Consciência , Eletroencefalografia , Reconhecimento Visual de Modelos , Humanos , Feminino , Masculino , Eletroencefalografia/métodos , Adulto , Estado de Consciência/fisiologia , Adulto Jovem , Reconhecimento Visual de Modelos/fisiologia , Encéfalo/fisiologia , Encéfalo/diagnóstico por imagem , Reconhecimento Psicológico/fisiologia , Estimulação Luminosa/métodosRESUMO
Scoliosis, characterized by spine deformity, is most common in adolescent idiopathic scoliosis (AIS). Manual Cobb angle measurement limitations underscore the need for automated tools. This study employed a vertebral landmark extraction method and Feedforward Neural Network (FNN) to predict scoliosis progression in 79 AIS patients. The novel intervertebral angles matrix format showcased results. The mean absolute error for the intervertebral angle progression was 1.5 degrees, while the Pearson correlation of the predicted Cobb angles was 0.86. The accuracy in classifying Cobb angles (<15°, 15-25°, 25-35°, 35-45°, >45°) was 0.85, with 0.65 sensitivity and 0.91 specificity. The FNN demonstrated superior accuracy, sensitivity, and specificity, aiding in tailored treatments for potential scoliosis progression. Addressing FNNs' over-fitting issue through strategies like "dropout" or regularization could further enhance their performance. This study presents a promising step towards automated scoliosis diagnosis and prognosis.
RESUMO
Bipedalism is unique among mammals. Until modern times, a fall and resulting leg fracture could be fatal. Balance maintenance after a destabilizing event requires instantaneous decision making. The vestibular system plays an essential role in this process, initiating an emergency response. The afferent otolithic neural response is the first directionally oriented information to reach the cortex, and it can then be used to initiate an appropriate protective response. Some vestibular efferent axons feed directly into type I vestibular hair cells. This allows for rapid vestibular feedback via the striated organelle (STO), which has been largely ignored in most texts. We propose that this structure is essential in emergency fall prevention, and also that the system of sensory detection and resultant motor response works by having efferent movement information simultaneously transmitted to the maculae with the movement commands. This results in the otolithic membrane positioning itself precisely for the planned movement, and any error is due to an unexpected external cause. Error is fed back via the vestibular afferent system. The efferent system causes macular otolithic membrane movement through the STO, which occurs simultaneously with the initiating motor command. As a result, no vestibular afferent activity occurs unless an error must be dealt with.
RESUMO
Parkinson's Disease (PD) is a prevalent neurological condition characterized by motor and cognitive impairments, typically manifesting around the age of 50 and presenting symptoms such as gait difficulties and speech impairments. Although a cure remains elusive, symptom management through medication is possible. Timely detection is pivotal for effective disease management. In this study, we leverage Machine Learning (ML) and Deep Learning (DL) techniques, specifically K-Nearest Neighbor (KNN) and Feed-forward Neural Network (FNN) models, to differentiate between individuals with PD and healthy individuals based on voice signal characteristics. Our dataset, sourced from the University of California at Irvine (UCI), comprises 195 voice recordings collected from 31 patients. To optimize model performance, we employ various strategies including Synthetic Minority Over-sampling Technique (SMOTE) for addressing class imbalance, Feature Selection to identify the most relevant features, and hyperparameter tuning using RandomizedSearchCV. Our experimentation reveals that the FNN and KSVM models, trained on an 80-20 split of the dataset for training and testing respectively, yield the most promising results. The FNN model achieves an impressive overall accuracy of 99.11%, with 98.78% recall, 99.96% precision, and a 99.23% f1-score. Similarly, the KSVM model demonstrates strong performance with an overall accuracy of 95.89%, recall of 96.88%, precision of 98.71%, and an f1-score of 97.62%. Overall, our study showcases the efficacy of ML and DL techniques in accurately identifying PD from voice signals, underscoring the potential for these approaches to contribute significantly to early diagnosis and intervention strategies for Parkinson's Disease.
Assuntos
Aprendizado de Máquina , Doença de Parkinson , Doença de Parkinson/diagnóstico , Humanos , Masculino , Feminino , Pessoa de Meia-Idade , Idoso , Redes Neurais de Computação , Voz , Aprendizado ProfundoRESUMO
Individuals with untreated, mild-to-moderate recurrent neck pain or stiffness (subclinical neck pain (SCNP)) have been shown to have impairments in upper limb proprioception, and altered cerebellar processing. It is probable that aiming trajectories will be impacted since individuals with SCNP cannot rely on accurate proprioceptive feedback or feedforward processing (body schema) for movement planning and execution, due to altered afferent input from the neck. SCNP participants may thus rely more on visual feedback, to accommodate for impaired cerebellar processing. This quasi-experimental study sought to determine whether upper limb kinematics and oculomotor processes were impacted in those with SCNP. 25 SCNP and 25 control participants who were right-hand dominant performed bidirectional aiming movements using two different weighted styli (light or heavy) while wearing an eye-tracking device. Those with SCNP had a greater time to and time after peak velocity, which corresponded with a longer upper limb movement and reaction time, seen as greater constant error, less undershoot in the upwards direction and greater undershoot in the downwards direction compared to controls. SCNP participants also showed a trend towards a quicker ocular reaction and movement time compared to controls, while the movement distance was fairly similar between groups. This study indicates that SCNP alters aiming performances, with greater reliance on visual feedback, likely due to altered proprioceptive input leading to altered cerebellar processing.
Assuntos
Movimentos Oculares , Retroalimentação Sensorial , Cervicalgia , Propriocepção , Desempenho Psicomotor , Extremidade Superior , Humanos , Masculino , Feminino , Adulto , Fenômenos Biomecânicos/fisiologia , Propriocepção/fisiologia , Desempenho Psicomotor/fisiologia , Cervicalgia/fisiopatologia , Movimentos Oculares/fisiologia , Extremidade Superior/fisiopatologia , Mãos/fisiopatologia , Mãos/fisiologia , Tempo de Reação , Adulto Jovem , Tecnologia de Rastreamento Ocular , Movimento/fisiologia , Objetivos , Pessoa de Meia-IdadeRESUMO
The objective of this research is to examine the thermophysical features of magnetic parameter (Ha) and time step (τ) in a lid-driven cavity using a water-based Al2O3 nanofluid and the efficacy of ANN models in accurately predicting the average heat transfer rate. The Galerkin weighted residual approach is used to solve a set of dimensionless nonlinear governing equations. The Levenberg-Marquardt back propagation technique is used for training ANN using sparse simulated data. The findings of the investigation about the flow and thermal fields are shown. Furthermore, a comparative study and prediction have been conducted on the impact of manipulating factors on the average Nusselt number derived from the numerical heat transfer analysis. The findings of the research indicate that, in the absence of magnetohydrodynamics, a rise in the Hartmann number resulted in a drop in both the fluid velocity profile and magnitude. Conversely, it was observed that the temperature and Nusselt number exhibited an increase under these conditions. The mean temperature of the fluid rises as the Hartmann number drops, reaching a peak value of 0.114 when Ha = 0. The scenario where Ha = 0, representing the lack of magnetohydrodynamics, shows the highest average Nusselt number, whereas the instance with Ha = 45 presents the lowest Nusselt number. The ANN model has a high level of accuracy, as seen by an MSE value of 0.00069 and a MAE value of 0.0175, resulting in a 99% accuracy rate.
RESUMO
Electromagnetic actuation can support many fields of technology, such as robotics or biomedical applications. In this context, fully understanding the system behavior and proposing a low-cost package for feedback control is challenging. Modeling the electromagnetic force is particularly tricky because it is a nonlinear function of the actuated object's position and coil's current. Measuring in real time the position of the actuated object with the precision required for accurate motion control is also nontrivial. In this study, we propose a novel, cost-effective electromagnetic set-up to achieve position control via visual feedback. We actuated vertically and under different experimental conditions a 10 mm diameter steel ball hanging on a low-stiffness spring, demonstrating good tracking performance (the position error remained within ±0.5 mm, with a negligible phase delay in the best scenarios). The experimental results confirm the feasibility of the proposed set-up, which is characterized by minimum complexity and realized with off-the-shelf and cost-effective components. For these reasons, such a contribution helps to understand and apply electromagnetic actuation even further.
RESUMO
OBJECTIVE: To investigate the efficacy of a feedforward control-based intervention strategy for preventing hypothermia among trauma patients during pre-hospital emergency care. METHODS: We conducted a retrospective analysis comparing trauma patients treated before and after implementing the intervention, with 40 cases in each group. All patients received emergency care from the Fuzhou Emergency Center on the scene. Multivariate analysis was used to explore the risk factors for hypothermia. The effective rate, incidence of adverse reactions, quality of body temperature management, medical staff's knowledge, attitudes, and behaviors regarding mild hypothermia prevention, coagulation function, treatment time at various stages, prognosis score, and treatment situation were compared between the two groups. RESULTS: The adverse reactions, intervention methods, and degree of cognitive improvement were influencing factors for hypothermia. The effective rate (92.50%) in the feedforward control group was higher than that in the non-feedforward control group (65.00%), with a lower incidence of adverse reactions (2.50%). The temperature management quality score of the feedforward control group (6.23±0.62) was higher. The feedforward control group achieved a higher quality score for temperature management (6.23±0.62) and exhibited a greater understanding of hypothermia prevention among trauma patients (P<0.05). Compared to the non-feedforward control group, the feedforward control group showed improved coagulation function, better performance in treatment time at each node, and higher prognosis scores. CONCLUSION: The intervention model based on feedforward control can effectively improve the standard of pre-hospital emergency care and prevent the incidence of hypothermia in trauma patients.
RESUMO
Suspended sediment concentration prediction is critical for the design of reservoirs, dams, rivers ecosystems, various operations of aquatic resource structure, environmental safety, and water management. In this study, two different machine models, namely the cascade correlation neural network (CCNN) and feedforward neural network (FFNN) were applied to predict daily-suspended sediment concentration (SSC) at Simga and Jondhara stations in Sheonath basin, India. Daily-suspended sediment concentration and discharge data from 2010 to 2015 were collected and used to develop the model to predict suspended sediment concentration. The developed models were evaluated using statistical indices like Nash and Sutcliffe efficiency coefficient (NES), root mean square error (RMSE), Willmott's index of agreement (WI), and Legates-McCabe's index (LM), supplemented by a scatter plot, density plots, histograms and Taylor diagram for graphical representation. The developed model was evaluated and compared with CCNN and FFNN. Nine input combinations were explored using different lag-times for discharge (Qt-n) and suspended sediment concentration (St-n) as input variables, with the current suspended sediment concentration as the desired output, to develop CCNN and FFNN models. The CCNN4 model with 4 lagged inputs (St-1, St-2, St-3, St-4) outperformed the other developed models with the lowest RMSE = 95.02 mg/l and the highest NES = 0.0.662, WI = 0.890 and LM = 0.668 for the Jondhara Station while the same CCNN4 model secure as the best with the lowest RMSE = 53.71 mg/l and the highest NES = 0.785, WI = 0.936 and LM = 0.788 for the Simga Station. The result shows the CCNN model was better than the FFNN model for predicting daily-suspended sediment at both stations in the Sheonath basin, India. Overall, CCNN showed better forecasting potential for suspended sediment concentration compared to FFNN at both stations, demonstrating their applicability for hydrological forecasting with complex relationships.
RESUMO
This paper studies a multi-hydraulic system (MHS) synchronization control algorithm. Firstly, a general nonlinear asymmetric MHS state space entirety model is established and subsequently the model form is simplified by nonlinear feedback linearization. Secondly, an entirety model-type solution is proposed, integrating a nonlinear model predictive control (NMPC) algorithm with a cross-coupling control (CCC) algorithm. Furthermore, a novel disturbance compensator based on the system's inverse model is introduced to effectively handle disturbances, encompassing unmodeled errors and noise. The proposed innovative controller, known as nonlinear model predictive control-cross-coupling control with deep neural network feedforward (NMPC-CCC-DNNF), is designed to minimize synchronization errors and counteract the impact of disturbances. The stability of the control system is rigorously demonstrated. Finally, simulation results underscore the efficacy of the NMPC-CCC-DNNF controller, showcasing a remarkable 60.8% reduction in synchronization root mean square error (RMSE) compared to other controllers, reaching up to 91.1% in various simulations. These results affirm the superior control performance achieved by the NMPC-CCC-DNNF controller.
RESUMO
Appropriate weight initialization settings, along with the ReLU activation function, have become cornerstones of modern deep learning, enabling the training and deployment of highly effective and efficient neural network models across diverse areas of artificial intelligence. The problem of "dying ReLU," where ReLU neurons become inactive and yield zero output, presents a significant challenge in the training of deep neural networks with ReLU activation function. Theoretical research and various methods have been introduced to address the problem. However, even with these methods and research, training remains challenging for extremely deep and narrow feedforward networks with ReLU activation function. In this paper, we propose a novel weight initialization method to address this issue. We establish several properties of our initial weight matrix and demonstrate how these properties enable the effective propagation of signal vectors. Through a series of experiments and comparisons with existing methods, we demonstrate the effectiveness of the novel initialization method.