Taste sensation can be objectively measured using electroencephalography (EEG) or electromyography (EMG). How-ever, it is still challenging to effectively utilize the complementary information from EEG and EMG signals in taste sensation recognition. This paper proposes a bimodal fusion network (Bi-FusionNet) for recognizing basic taste sensations (sour, sweet, bitter, salty, umami, and blank). Two convolutional backbones with similar structures are designed to separately extract the single-modal features of EEG and EMG. Then, EEG and EMG features are concatenated for bimodal interaction and complementarity. Finally, three loss functions are adopted: a center loss for aggregating intra-class samples, a mean squared error loss for sequence positions for minimizing the difference between signals during the stimulation, and a softmax loss for minimizing the entropy of prediction and true labels. The results on the taste sensation dataset show that bimodal fusion improves recognition performance, and Bi-FusionNet outperforms single-modal methods and other fusion methods. Bi-FusionNet paves the way for the application of multimodal fusion in taste sensation recognition.

In this article, the consensus of constrained linear heterogeneous multiagent systems under prediction and optimization is investigated. By optimizing the consensus problems constrained to state equations and general linear constraints, two types of distributed analytical model predictive controllers are proposed. Furthermore, stability conditions for the two types of controllers are derived, in which a relationship between the communication topology and dynamics of heterogeneous agents is clarified. Simulation examples of networked heterogeneous agents illustrate the convergence and validity of the proposed controllers.

This work explores the effectiveness of the intensity-varied closed-loop noise stimulation on the oscillation suppression in the Parkinsonian state. Deep brain stimulation (DBS) is the standard therapy for Parkinson's disease (PD), but its effects need to be improved. The noise stimulation has compelling results in alleviating the PD state. However, in the open-loop control scheme, the noise stimulation parameters cannot be self-adjusted to adapt to the amplitude of the synchronized neuronal activities in real time. Thus, based on the delayed-feedback control algorithm, an intensity-varied closed-loop noise stimulation strategy is proposed. Based on a computational model of the basal ganglia (BG) that can present the intrinsic properties of the BG neurons and their interactions with the thalamic neurons, the proposed stimulation strategy is tested. Simulation results show that the noise stimulation suppresses the pathological beta (12-35 Hz) oscillations without any new rhythms in other bands compared with traditional high-frequency DBS. The intensity-varied closed-loop noise stimulation has a more profound role in removing the pathological beta oscillations and improving the thalamic reliability than open-loop noise stimulation, especially for different PD states. And the closed-loop noise stimulation enlarges the parameter space of the delayed-feedback control algorithm due to the randomness of noise signals. We also provide a theoretical analysis of the effective parameter domain of the delayed-feedback control algorithm by simplifying the BG model to an oscillator model. This exploration may guide a new approach to treating PD by optimizing the noise-induced improvement of the BG dysfunction.

The further exploration of the neural mechanisms underlying the biological activities of the human brain depends on the development of large-scale spiking neural networks (SNNs) with different categories at different levels, as well as the corresponding computing platforms. Neuromorphic engineering provides approaches to high-performance biologically plausible computational paradigms inspired by neural systems. In this article, we present a biological-inspired cognitive supercomputing system (BiCoSS) that integrates multiple granules (GRs) of SNNs to realize a hybrid compatible neuromorphic platform. A scalable hierarchical heterogeneous multicore architecture is presented, and a synergistic routing scheme for hybrid neural information is proposed. The BiCoSS system can accommodate different levels of GRs and biological plausibility of SNN models in an efficient and scalable manner. Over four million neurons can be realized on BiCoSS with a power efficiency of 2.8k larger than the GPU platform, and the average latency of BiCoSS is 3.62 and 2.49 times higher than conventional architectures of digital neuromorphic systems. For the verification, BiCoSS is used to replicate various biological cognitive activities, including motor learning, action selection, context-dependent learning, and movement disorders. Comprehensively considering the programmability, biological plausibility, learning capability, computational power, and scalability, BiCoSS is shown to outperform the alternative state-of-the-art works for large-scale SNN, while its real-time computational capability enables a wide range of potential applications.

In the face of a wide variety and a large number of production and domestic waste, it is a great challenge for the task of automatic detection and sorting of waste. Based on yolov5 algorithm, this paper proposes a method for rapid detection and classification of garbage, trains the model on taco[1] garbage data set, and extracts the location and feature information of garbage through this network model according to the experimental results. In reality, this model can effectively detect the garbage classified by the data set. After testing, the mAP(Mean Average Percision) value of the model reaches 97.62%, the detection accuracy is 95.49%, and the detection speed reaches 5.52fps. Compared with yolov3 network model, which better complete the task of garbage classification and detection. This network model has the necessary technical conditions for the algorithm of waste sorting robots.

Multicompartment emulation is an essential step to enhance the biological realism of neuromorphic systems and to further understand the computational power of neurons. In this paper, we present a hardware efficient, scalable, and real-time computing strategy for the implementation of large-scale biologically meaningful neural networks with one million multi-compartment neurons (CMNs). The hardware platform uses four Altera Stratix III field-programmable gate arrays, and both the cellular and the network levels are considered, which provides an efficient implementation of a large-scale spiking neural network with biophysically plausible dynamics. At the cellular level, a cost-efficient multi-CMN model is presented, which can reproduce the detailed neuronal dynamics with representative neuronal morphology. A set of efficient neuromorphic techniques for single-CMN implementation are presented with all the hardware cost of memory and multiplier resources removed and with hardware performance of computational speed enhanced by 56.59% in comparison with the classical digital implementation method. At the network level, a scalable network-on-chip (NoC) architecture is proposed with a novel routing algorithm to enhance the NoC performance including throughput and computational latency, leading to higher computational efficiency and capability in comparison with state-of-the-art projects. The experimental results demonstrate that the proposed work can provide an efficient model and architecture for large-scale biologically meaningful networks, while the hardware synthesis results demonstrate low area utilization and high computational speed that supports the scalability of the approach.

The benefit of noise in improving the basal ganglia (BG) dysfunctions, especially Parkinsonian state, is explored in this paper. High frequency (≥ 100 Hz) deep brain stimulation (DBS), as a clinical effective stimulation method, has compelling and fantastic results in alleviating the motor symptoms of Parkinson's disease (PD). However, the mechanism of DBS is still unclear. And the selection of the DBS waveform parameters faces great challenges to further optimize the stimulation effects and to reduce its energy expenditure. Considering that the desynchronization of the BG neuronal activities is benefited from the forced high frequency regular spikes driven by standard high frequency DBS, we expect to explore a novel stimulation method that has capability of restoring the BG physiological firing patterns without introducing artificial high-frequency fires. In this paper, a colored noise stimulation is used as a neuromodulation method to disrupt the firing patterns of the pathological neuronal activities. A computational model of the BG that exhibits the intrinsic properties of the BG neurons and their interactions with the thalamic (Th) cells is employed. Based on the model, we investigate the effects of noise stimulation and explore the impacts of the noise stimulation parameters on both relay reliability of the Th neurons and energy expenditure of the stimulation. By comparison, it can be found that noise stimulation does not entrain the network to an artificial high-frequency firing state, but induces the pathological increased synchronous activities back to a normal physiological level. Moreover, besides the capability of restoring the neuronal state, the benefits of the noise also include its balanced waveform to avert potential tissue or electrode damage and its ability to reduce the energy expenditure to 50% less than that of the standard DBS, when the noise stimulation has low frequency (≤ 100 Hz) and appropriate intensity. Thus, the exploration of the optimal noise-induced improvement of the BG dysfunction is of great significance in treating symptoms of neurological disorders such as PD.

The investigation of the human intelligence, cognitive systems and functional complexity of human brain is significantly facilitated by high-performance computational platforms. In this paper, we present a real-time digital neuromorphic system for the simulation of large-scale conductance-based spiking neural networks (LaCSNN), which has the advantages of both high biological realism and large network scale. Using this system, a detailed large-scale cortico-basal ganglia-thalamocortical loop is simulated using a scalable 3-D network-on-chip (NoC) topology with six Altera Stratix III field-programmable gate arrays simulate 1 million neurons. Novel router architecture is presented to deal with the communication of multiple data flows in the multinuclei neural network, which has not been solved in previous NoC studies. At the single neuron level, cost-efficient conductance-based neuron models are proposed, resulting in the average utilization of 95% less memory resources and 100% less DSP resources for multiplier-less realization, which is the foundation of the large-scale realization. An analysis of the modified models is conducted, including investigation of bifurcation behaviors and ionic dynamics, demonstrating the required range of dynamics with a more reduced resource cost. The proposed LaCSNN system is shown to outperform the alternative state-of-the-art approaches previously used to implement the large-scale spiking neural network, and enables a broad range of potential applications due to its real-time computational power.

One of the challenging problems in real-time control of movement disorders is the effective handling of time-variant brain activities that involve stochastic functional networks with nonlinear dynamics. For such challenges in neuromodulation tasks, fuzzy logic control (FLC) has shown significant potential. The objective of this paper is to present a FLC-based strategy to treat pathological symptoms of movement-disorder with higher performance. The strategy is two-fold: first, develop a design methodology for the FLC system that can robustly control pathological conditions and significantly improve control performance; and second, develop a hardware-efficient implementation for real-time neuromodulation applications. To enhance control performance, a hidden variable in the neural network that can be estimated using an unscented Kalman filter is identified as a feedback variable. In comparison with state-of-the-art schemes, the proposed design can adaptively optimize the control signals without requiring particular information of the controlled plant, thus avoiding repeated determinations of controller parameters. A field-programmable gate array is used for the reconfigurable realization of the entire control strategy based on a modification of the original neural network. The presented design, with enhanced control performance and higher hardware efficiency, has significant potential for clinical treatment of movement disorders and offers a new perspective on applications in the fields of neural control engineering and brain-machine interfaces.

Enhanced beta (12-30 Hz) oscillatory activity in the basal ganglia (BG) is a prominent feature of the Parkinsonian state in animal models and in patients with Parkinson's disease. Increased beta oscillations are associated with severe dopaminergic striatal depletion. However, the mechanisms underlying these pathological beta oscillations remain elusive. Inspired by the experimental observation that only subsets of neurons within each nucleus in the BG exhibit oscillatory activities, a computational model of the BG-thalamus neuronal network is proposed, which is characterized by subdivided nuclei within the BG. Using different currents externally applied to the neurons within a given nucleus, neurons behave according to one of the two subgroups, named "-N" and "-P," where "-N" and "-P" denote the normal and the Parkinsonian states, respectively. The ratio of "-P" to "-N" neurons indicates the degree of the Parkinsonian state. Simulation results show that if "-P" neurons have a high degree of connectivity in the subthalamic nucleus (STN), they will have a significant downstream effect on the generation of beta oscillations in the globus pallidus. Interestingly, however, the generation of beta oscillations in the STN is independent of the selection of the "-P" neurons in the external segment of the globus pallidus (GPe), despite the reciprocal structure between STN and GPe. This computational model may pave the way to revealing the mechanism of such pathological behaviors in a realistic way that can replicate experimental observations. The simulation results suggest that the STN is more suitable than GPe as a deep brain stimulation target.

This paper designs a ground-target tracking system with unmanned aerial vehicles(UAVs), which consists of the states estimation of the ground target and the distributed model predictive control with two UAVs. In the tracking system, the extended kalman filter estimates the states of the target, and the distributed model predictive control in the UAVs is responsible to track the target and avoid collision. Both the fixed-wing and rotorcraft UAVs are successfully realized. Numerical simulations demonstrate the effectiveness of our tracking systems with the designed UAVs.

A generalized predictive closed-loop control strategy to improve the basal ganglia activity patterns in Parkinson's disease (PD) is explored in this paper. Based on system identification, an input-output model is established to reveal the relationship between external stimulation and neuronal responses. The model contributes to the implementation of the generalized predictive control (GPC) algorithm that generates the optimal stimulation waveform to modulate the activities of neuronal nuclei. By analyzing the roles of two critical control parameters within the GPC law, optimal closed-loop control that has the capability of restoring the normal relay reliability of the thalamus with the least stimulation energy expenditure can be achieved. In comparison with open-loop deep brain stimulation and traditional static control schemes, the generalized predictive closed-loop control strategy can optimize the stimulation waveform without requiring any particular knowledge of the physiological properties of the system. This type of closed-loop control strategy generates an adaptive stimulation waveform with low energy expenditure with the potential to improve the treatments for PD.

In this letter, a novel hierarchical sparse representation-based classification (HSRC) for synthetic aperture radar (SAR) images is proposed. Features utilized in HSRC are extracted from the multisize patches around each pixel to precisely describe the complex terrains. Two thresholds are introduced in the sparse representation classifier to restrict the range of reconstruction residual, which classifies the reliable classified points, and the rest of the pixels are considered as the uncertain ones in the original SAR image. Then, a new dictionary is constructed by the reliable pixels, and the uncertain pixels will be reclassified in the next classification layer. The hierarchical structure is very reasonable and effective to employ simple features in each layer for describing the various topographic types. Compared with traditional sparse representation-based classification and support vector machines in several fixed-size patches, the proposed method can obtain better performance both in quantitative evaluation and visualization results.

The objective here is to explore the use of adaptive input-output feedback linearization method to achieve an improved deep brain stimulation (DBS) algorithm for closed-loop control of Parkinson's state. The control law is based on a highly nonlinear computational model of Parkinson's disease (PD) with unknown parameters. The restoration of thalamic relay reliability is formulated as the desired outcome of the adaptive control methodology, and the DBS waveform is the control input. The control input is adjusted in real time according to estimates of unknown parameters as well as the feedback signal. Simulation results show that the proposed adaptive control algorithm succeeds in restoring the relay reliability of the thalamus, and at the same time achieves accurate estimation of unknown parameters. Our findings point to the potential value of adaptive control approach that could be used to regulate DBS waveform in more effective treatment of PD.

A novel closed-loop control strategy is proposed to control Parkinsonian state based on a computational model. By modeling thalamocortical relay neurons under external electric field, a slow variable feedback control is applied to restore its relay functionality. Qualitative and quantitative analysis demonstrates the performance of feedback controller based on slow variable is more efficient compared with traditional feedback control based on fast variable. These findings point to the potential value of model-based design of feedback controllers for Parkinson's disease.