Biological Nervous Systems Research Articles

Oxygen-Detecting Synaptic Device for Realization of Artificial Autonomic Nervous System for Maintaining Oxygen Homeostasis.

Incorporation of various functions of a biological nervous system into electronic devices is an intriguing challenge in the realization of a human-like recognition and response system. Emerging artificial synaptic devices capable of processing electronic signals through neuromorphic functions operate such biomimetic systems similarly to biological nervous systems. Here, an oxygen-sensitive artificial synaptic device that simultaneously detects oxygen concentration and generates a synaptic signal is demonstrated. The device successfully achieves an interconversion between the excitatory and inhibitory modes of the synaptic current at various oxygen concentrations by virtue of an oxygen-sensitive trilayered organic double heterojunction. The oxygen-induced traps in the organic layer modulate the majority charge carrier from holes to electrons, and this modulation induces an interconversion between the excitatory and inhibitory modes according to the environmental oxygen condition. Finally, the proposed synaptic device is applied to the realization of a negative feedback system for regulation of oxygen homeostasis, which mimics the human autonomic nervous system. The oxygen-sensitive synaptic device proposed in this study is expected to open up new possibilities for the development of a biomimetic neural system that can respond appropriately to various environmental changes.

On-Chip Training Spiking Neural Networks Using Approximated Backpropagation With Analog Synaptic Devices.

Hardware-based spiking neural networks (SNNs) inspired by a biological nervous system are regarded as an innovative computing system with very low power consumption and massively parallel operation. To train SNNs with supervision, we propose an efficient on-chip training scheme approximating backpropagation algorithm suitable for hardware implementation. We show that the accuracy of the proposed scheme for SNNs is close to that of conventional artificial neural networks (ANNs) by using the stochastic characteristics of neurons. In a hardware configuration, gated Schottky diodes (GSDs) are used as synaptic devices, which have a saturated current with respect to the input voltage. We design the SNN system by using the proposed on-chip training scheme with the GSDs, which can update their conductance in parallel to speed up the overall system. The performance of the on-chip training SNN system is validated through MNIST data set classification based on network size and total time step. The SNN systems achieve accuracy of 97.83% with 1 hidden layer and 98.44% with 4 hidden layers in fully connected neural networks. We then evaluate the effect of non-linearity and asymmetry of conductance response for long-term potentiation (LTP) and long-term depression (LTD) on the performance of the on-chip training SNN system. In addition, the impact of device variations on the performance of the on-chip training SNN system is evaluated.

How we created neuromorphic engineering

Neuromorphic engineering aims to create computing hardware that mimics biological nervous systems, and it is expected to play a key role in the next era of hardware development. Carver Mead recounts how it all began.

A Comprehensive Study on Architecture of Neural Networks and Its Prospects in Cognitive Computing

This paper proffers an overview of neural network, coupled with early neural network architecture, learning methods, and applications. Basically, neural networks are simplified models of biological nervous systems and that's why they have drawn crucial attention of research community in the domain of artificial intelligence. Basically, such networks are highly interconnected networks possessing a huge number of processing elements known as neurons. Such networks learn by examples and exhibit the mapping capabilities, generalization, fault resilience conjointly with escalated rate of information processing. In the current paper, various types of learning methods employed in case of neural networks are discussed. Subsequently, the paper details the deep neural network (DNN), its key concepts, optimization strategies, activation functions used. Afterwards, logistic regression and conventional optimization approaches are described in the paper. Finally, various applications of neural networks in various domains are included in the paper before concluding it.

Canard-induced mixed-mode oscillations and bifurcation analysis in a reduced 3D pyramidal cell model

This study entails investigation of mixed-mode oscillations (MMOs) with a conductance-based pyramidal cell (PC) model located in the entorhinal cortex layer V. This six-dimensional neuron model was reduced to three dimensions by analysis of the voltage-dependent timescales to illustrate a regime in which the MMOs are generated. Additionally, the mechanism of generation of the MMOs under antiepileptic drug conditions was illustrated in the 3D PC model. Combined with geometric singular perturbation theory (GSPT), this work shows that there is a range of parameters under which the reduced model explains the emergence of MMOs caused by an underlying canard phenomenon. In particular, we theoretically calculate the number of subthreshold oscillations using the relationship with the eigenvalue ratio of the singular perturbation system at the folded singular node, which is consistent with numerical simulations. Furthermore, a slow–fast dynamics analysis of the 3D PC model is performed, where two slow/one fast and one slow/two fast systems with the layer problem and the reduced problem are considered to explain the trajectory on the critical manifold. General one- and two-parameter bifurcation types are also discussed in this work. The first Lyapunov coefficient of the Hopf bifurcation can decide whether the bifurcation is supercritical or subcritical. Bogdanov–Takens (BT) bifurcation was also analyzed in this study and associated with three bifurcation curves near the BT point. Finally, studies on the GSPT and bifurcation analysis are of great importance for further understanding the complex dynamic behaviors and crucial roles of the signal transmission and information processing pathways of the biological nervous system.

Analog based Neuromorphic Systems on Low Power Current Mode Circuits

Neuromorphic computing is the process used to appliance the neural system models. Formerly, it is referred to as the biological process and later it turned out to be the computing algorithms. Many neuromorphic algorithms represented as the neural figures such as neural spikes, fluctuated graphs, and synapses. The biological nervous system for instance consists of huge number of neurons and they collectively work to encode the stimulus of various senses. In case of neuromorphic computing, automated brain brings in the concept of efficient work carried out through artificial means. The neuromorphic computing thus evolves as a major technological advancement and the need of such technique is the need of the hour in various scientific as well as field applications. In existing techniques, the scaling, power and area are not efficient. This study attempts to address the major issues such as scaling and power. This paper explains the design on a non-spiking network which is used for population coding architecture. The model which is discussed in this paper is based on the analog domain and the current mode circuits are also involved. The input neuron model consists of current direction selector block, current scale block and minimum current block which all comprise to form the neuron model. This paper also brings out the possible outcome of low power constraints. This paper involves 180nm technology with which the power is measured. This paper brings out the simulations of both 180 and 90nm technologies. Apart from current scale block, minimum current block and current direction selector block, there are other blocks such as current splitter block and current mode low pass filter block, where all the circuits work under the sub-threshold condition. The power consumption obtained in the 180 nm technology is 58.838 μW and its energy equivalent is 1.765pJ. Neuromorphic computing is used as an application where the machines are being automated and such machines come with self-thinking capability. Neuromorphic computing design which is evolved from this paper is found to be more power ad energy efficient. The tool used is Cadence Virtuoso.

Chaotic dynamics in a neural network with different types of external stimuli

Biological nervous system is very sensitive to external disturbances, and appropriate stimulus is beneficial for improving neural function in the neural system. In this paper, the effect of different external stimuli on chaotic dynamics in a Hopfield neural network with three neurons is explored. Mathematical model of the neural network is respectively established under three different cases, namely without external stimulus, with only electromagnetic radiation stimulus, and with both electromagnetic radiation stimulus and multi-level-logic pulse stimulus. Under the three cases, equilibrium points, stabilities, and attractors of the neural network are investigated carefully. The research results demonstrate that the neural network with periodic attractors can induce abundant chaotic attractors by imposing electromagnetic radiation on its one neuron. And when this neuron is simultaneously stimulated via electromagnetic radiation and multi-level-logic pulse, the neural network can produce complex multi-scroll attractors previously unobserved in Hopfield-type neural networks. Numerical results are verified by hardware experiments, effectively. Furthermore, based on the Helmholtzâs theorem, the Hamilton energy of the neural network is calculated and analyzed. It is found that lower average Hamilton energy can be detected in the neural network when complexity of external stimuli is enhanced. These new findings could offer a new insight into the occurrence mechanism of some neurological diseases.

A Study on Brain Tumor Detection and Segmentation Using Deep Learning Techniques

It’s a new era technology in the field of medical engineering giving awareness about the various healthcare features. Deep learning is a part of machine learning, it is capable of handling high dimensional data and is efficient in concentrating on the right features. Tumor is an unbelievably complex disease: a multifaceted cell has more than hundred billion cells; each cell acquires mutation exclusively. Detection of tumor particles in experiment is easily done by MRI or CT. Brain tumors can also be detected by MRI, however, deep learning techniques give a better approach to segment the brain tumor images. Deep Learning models are imprecisely encouraged by information handling and communication designs in biological nervous system. Classification plays an significant role in brain tumor detection. Neural network is creating a well-organized rule for classification. To accomplish medical image data, neural network is trained to use the Convolution algorithm. Multilayer perceptron is intended for identification of a image. In this study article, the brain images are categorized into two types: normal and abnormal. This article emphasize the importance of classification and feature selection approach for predicting the brain tumor. This classification is done by machine learning techniques like Artificial Neural Networks, Support Vector Machine and Deep Neural Network. It could be noted that more than one technique can be applied for the segmentation of tumor. The several samples of brain tumor images are classified using deep learning algorithms, convolution neural network and multi-layer perceptron.

Multi-frequency sinusoidal chaotic neural network and its complex dynamics

A large number of animal experiments show that there is irregular chaos in the biological nervous systems. An artificial chaotic neural network is a highly nonlinear dynamic system, which can realize a series of complex dynamic behaviors, optimize global search and neural computation, and generate pseudo-random sequences for information encryption. According to the superposition theory of sinusoidal signals with different frequencies of brain waves, a non-monotone activation function based on the multifrequency-frequency conversion sinusoidal function and a piecewise function is proposed to make a neural network more consistent with the biological characteristics. The analysis shows that by adjusting the parameters, the activation function can exhibit the EEG signals in its different states, which can simulate the rich and varying brain activities when the brain waves of different frequencies and types work at the same time. According to the activation function we design a new chaotic cellular neural network. The complexity of the chaotic neural network is analyzed by the structural complexity based SE algorithm and C0 algorithm. By means of Lyapunov exponential spectrum, bifurcation diagram and basin of attraction, the effects of the activation function’s parameters on its dynamic characteristics are analyzed in detail, and it is found that a series of complex phenomena appears in the chaotic neural network, such as many different types of chaotic attractors, coexistent chaotic attractors and coexistence limit cycles, which improves the performance of the chaotic neural network, and proves that the multi-frequency sinusoidal chaotic neural network has rich dynamic characteristics, so it has a good prospect in information processing, information encryption and other aspects.

Hidden extreme multistability with hyperchaos and transient chaos in a Hopfield neural network affected by electromagnetic radiation

Biological nervous system function is closely related to its dynamical behaviors, and some dynamical phenomena observed in biological systems can be detected in the simplified neural models. In this paper, the chaotic dynamics in a three-neuron-based Hopfield neural network (HNN) with stimulation of electromagnetic radiation is investigated. The neural network is modeled by utilizing a flux-controlled memristor to describe the effects of electromagnetic field on neurons. The simple neural model affected by electromagnetic radiation does not contain any equilibrium points, but can induce coexisting infinitely many hidden attractors, such as hyperchaos, transient hyperchaos, period, quasi-period, chaos as well as transient chaos with different chaotic times. In particular, the dynamics of hidden extreme multistability with hyperchaos and transient chaos in the neural network highly depends on the system parameters and state initial values. The coexistence of multiple hidden attractors is revealed via applying a host of numerical analysis methods including phase plots, time sequence waveforms, bifurcation diagrams, Lyapunov exponents and attraction basins. Besides, a HNN-based circuit consisting of commercially available electronic elements is designed to verify the theoretical analysis. Hardware measurement and MULTISIM simulation results are basically consistent with MATLAB numerical simulation results.

Bioinspired Flexible, Dual‐Modulation Synaptic Transistors toward Artificial Visual Memory Systems

AbstractA prominent challenge for artificial synaptic devices toward artificial perception systems is hardware redundancy, which demands neuromorphic devices that integrate both sensing and processing functions. Inspired by the biological visual and nervous systems, a novel flexible, dual‐modulation synaptic field‐effect transistor (SFET) is demonstrated in this work. The flexible SFET is constructed with zinc oxide nanowires and sodium alginate, which acts as the semiconductor layer and the gate dielectric, respectively. An excitatory postsynaptic current in this artificial synapse can be triggered by both electrical and ultraviolet stimuli as presynaptic spikes as a result of the electric double layer effect and the photoelectric effect. More importantly, through the co‐modulation of light and electric stimuli, the memory level of the artificial synapses can be tuned based on the transformation between short‐term plasticity and long‐term plasticity initiated by the gate voltage. Different voltages can modulate the memory retention levels of the optical inputs similar to the function of the optic nerve system. The underlying mechanisms for the SFET are investigated using Fourier transform infrared spectroscopy, photoluminescence, and X‐ray photoelectron spectroscopy. Overall, the devices provide a novel idea to mimic visual memory, showing a promising strategy for future electronic eyes.

ACVR1 Function in Health and Disease.

Activin A receptor type I (ACVR1) encodes for a bone morphogenetic protein type I receptor of the TGFβ receptor superfamily. It is involved in a wide variety of biological processes, including bone, heart, cartilage, nervous, and reproductive system development and regulation. Moreover, ACVR1 has been extensively studied for its causal role in fibrodysplasia ossificans progressiva (FOP), a rare genetic disorder characterised by progressive heterotopic ossification. ACVR1 is linked to different pathologies, including cardiac malformations and alterations in the reproductive system. More recently, ACVR1 has been experimentally validated as a cancer driver gene in diffuse intrinsic pontine glioma (DIPG), a malignant childhood brainstem glioma, and its function is being studied in other cancer types. Here, we review ACVR1 receptor function and signalling in physiological and pathological processes and its regulation according to cell type and mutational status. Learning from different functions and alterations linked to ACVR1 is a key step in the development of interdisciplinary research towards the identification of novel treatments for these pathologies.

Neuromorphic Liquid Marbles with Aqueous Carbon Nanotube Cores.

Neuromorphic computing devices attempt to emulate features of biological nervous systems through mimicking the properties of synapses toward implementing the emergent properties of their counterparts, such as learning. Inspired by recent advances in the utilization of liquid marbles (LMs, microliter quantities of fluid coated in hydrophobic powder) for the creation of unconventional computing devices, we describe the development of LMs with neuromorphic properties through the use of copper coatings and 1.0 mg mL–1 carbon nanotube (CNT)-containing fluid cores. Experimentation was performed through sandwiching the LMs between two cup-style electrodes and stimulating them with repeated dc pulses at 3.0 V. Our results demonstrate that “entrainment” of CNT-filled copper LMs via periodic pulses can cause their electrical resistance to rapidly switch between high to low resistance profiles upon inverting the polarity of stimulation: the reduction in resistance between high and low profiles was approximately 88% after two rounds of entrainment. This effect was found to be reversible through reversion to the original stimulus polarity and was strengthened by repeated experimentation, as evidenced by a mean reduction in time to switching onset of 43%. These effects were not replicated in nanotube solutions not bound inside LMs. Our electrical characterization also reveals that nanotube-filled LMs exhibit pinched loop hysteresis IV profiles consistent with the description of memristors. We conclude by discussing the applications of this technology to the development of unconventional computing devices and the study of emergent characteristics in biological neural tissue.

Versatile neuromorphic electronics by modulating synaptic decay of single organic synaptic transistor: From artificial neural networks to neuro-prosthetics

Organic neuromorphic electronics are inspired by a biological nervous system. Bio-inspired computing mimics learning and memory in a brain (i.e., the central nervous system), and bio-inspired soft robotics and nervous prosthetics mimics the neural signal transmission of afferent/efferent nerves (i.e., the peripheral nervous system). Synaptic decay time of nerves differ among biological organs, so the decay time of artificial synapses should be tuned for their specific uses in neuro-inspired electronics. However, controlling a synaptic decay constant in a fixed synaptic device geometry for broad applications was not been achieved in previous research of neuromorphic electronic devices despite the importance to achieve broad applications from neuromorphic computing to neuro-prosthetics. Here, we tailored the synaptic decay constant of organic synaptic transistors with fixed materials and devices structure rather than changing the form of presynaptic spikes, which enabled broad applications from neuromorphic computing to neuro-prosthetics. To achieve this, the relation between crystallinity of the polymer semiconductor film and the synaptic decay constant was revealed. The crystallinity of the polymer controlled electrochemical-doping kinetics and resultant synaptic behaviors of artificial synaptic transistors. In this way, we demonstrated not only long-term retention for learning and memory that is useful for neuromorphic computing in ion-gel gated organic synaptic transistor (IGOST) but also the short-term retention for fast synaptic transmission that is useful for emulating peripheral nerves such as sensory and motor nerve. To prove the feasibility of our approach in a two different ways, we first simulated pattern recognition on the MNIST dataset of handwritten digits using an IGOST with long-term retention due to increased crystallinity and then, developed artificial auditory sensory nerves that combines an IGOST with short term retention due to disordered chain morphology in a polymer semiconductor, with a triboelectric acoustic sensor. We expect that our approach will provide a universal strategy to realize wide neuromorphic electronic applications.

Hybrid teaching–learning-based optimization and neural network algorithm for engineering design optimization problems

Neural network algorithm (NNA) is one of the newest meta-heuristic algorithms, which is inspired by biological nervous systems and artificial neural networks. Benefiting from the unique structure of artificial neural networks, NNA has good global search ability. However, slow convergence is its drawback that restricts its practical application. Teaching–learning-based optimization (TLBO) is an algorithm without any effort for fine tuning initial parameters, which has fast convergence speed while it is easy to fall into local optimum in solving complex global optimization problems. Considering the features of NNA and TLBO, an effective hybrid method based on TLBO and NNA, named TLNNA, is proposed for solving engineering optimization problems. The performance of TLNNA for 30 well-known unconstrained benchmark functions and 4 challenging engineering optimization problems is examined and the optimization results are compared with other competitive meta-heuristic algorithms. Such comparisons suggest that TLNNA has not only good global search ability of NNA but also fast convergence speed of TLBO and is more successful for most test problems in terms of solution quality and computational efficiency.

Towards a Quantum Field Theory for Optical Artificial Intelligence

Today, several socio-techno-economic drivers are steering the evolution of Telecommunications and Internet towards a growing exploitation of ultra-broadband infrastructures (e.g., 5G) and Artificial Intelligence (AI) systems. Focusing on the most promising AI technological approaches, Deep Neural Networks (DNNs) are outperforming in several applications domains. One of the possible explanations, elaborated in literature, is that DNN functioning is deeply rooted in the principles of theoretical Physics, specifically Quantum Field Theory (QFT) and Gauge theory. This is encouraging even more researches and experiments in the direction of a full exploitation of quantum computing and networking for the development of innovative Information Communication Technologies (ICT) and AI systems. In this innovation avenue, given that QFT and Gauge theory have been already proposed for modeling the brain and biological nervous systems, this paper explores the intriguing possibility of exploiting QFT principles also for future DNN, for instance by using electromagnetic waves effects in metamaterials. This appears to be a promising direction of future studies and experiments: therefore, the paper also describes the architecture of a simple optical DNN prototype, based on metamaterials, which is intended as a live test-bed, for simulations and experiments.

An Analysis of FPGA Hardware Platform Based Artificial Neural Network

The artificial neural networks (ANN) algorithm is a mathematical model of a network by applying neurons and usually, it is represented as a directed graph with vertexes and edges. This algorithm is a paradigm of information processing to describe arbitrary functions and inspired by the biological nervous systems. This algorithm can learn a non-linearly separated set of outputs. A much software-based application has implemented this algorithm but the challenge in the hardware-based is still possible to do in mapping the output into binary values. This paper addressed on the platform of the hardware-based the ANN algorithm. The field programmable gate arrays (FPGAs) is closely related to this platform. The FPGAs have achieved a successful penetration in different multi-disciplinary domains.

Artificial intelligence in surgical pathology

Artificial intelligence (AI) is machine intelligence that mimics human cognitive function. It denotes the intelligence presented by some artificial entities including computers and robots. In supervised learning, a machine is trained with data that contain pairs of inputs and outputs. In unsupervised learning, machines are given data inputs that are not explicitly programmed.1 Machine learning refines a model that predicts outputs using sample inputs (features) and a feedback loop. It relies heavily on extracting or selecting salient features, which is a combination of art and science (“feature engineering”). A subset of feature learning is deep learning, which harnesses neural networks modeled after the biological nervous system of animals. Deep learning discovers the features from the raw data provided during training. Hidden layers in the artificial neural network represent increasingly more complex features in the data. Convolutional neural network is a type of deep learning commonly used for image analysis.

Organic Synapses for Neuromorphic Electronics: From Brain-Inspired Computing to Sensorimotor Nervetronics.

Living organisms have a long evolutionary history that has provided them with functions and structures that enable them to survive in their environment. The goal of biomimetic technology is to emulate these traits of living things. Research in bioinspired electronics develops electronic sensors and motor systems that mimic biological sensory organs and motor systems and that are intended to be used in bioinspired applications such as humanoid robots, exoskeletons, and other devices that combine a living body and an electronic device. To develop bioinspired robotic and electronic devices that are compatible with the living body at the neuronal level and that are operated by mechanisms similar to those in a living body, researchers must develop biomimetic electronic sensors, motor systems, brains, and nerves. Artificial organic synapses have emulated the brain's plasticity with much simpler structures and lower fabrication cost than neurons based on silicon circuits, and with smaller energy consumption than traditional von Neumann computing methods. Organic synapses are promising components of future neuromorphic systems. In this Account, we review recent research trends of neuromorphic systems based on organic synapses, then suggest research directions. We introduce the device structures and working mechanisms of reported organic synapses and the brain's plasticity, which are mainly imitated to demonstrate the learning and memory function of the organic synapses. We also introduce recent reports on sensory synapses and sensorimotor nervetronics that mimic biological sensory and motor nervous systems. Sensory nervetronics can be used to augment the sensory functions of the living body and to comprise the sensory systems of biomimetic robots. Organic synapses can also be used to control biological muscles and artificial muscles that have the same working mechanism as biological muscle. Motor nervetronics would impart life-like motion to bioinspired robots. Chemical approaches may provide insights to guide development of new organic materials, device structures, and working mechanisms to improve synaptic responses of organic neuromorphic systems. For example, organic synapses can be applied to electronic and robotic skins and bioimplantable medical devices that use mechanically stable, self-healing, and biocompatible organic materials. Biochemical approaches may expand the plasticity of the brain and nervous system. We expect that organic neuromorphic systems will be vital components in bioinspired robotic and electronic applications, including biocompatible neural prosthetics, exoskeletons, humanoid soft robots, and cybernetics devices that are integrated with biological and artificial organs.

Bio-inspired sensing and actuating architectures for feedback control of civil structures

Civil structures, such as buildings and bridges, are constantly at risk of failure due to external environmental loads, such as earthquakes or strong winds. To minimize the effects of these loads, active feedback control systems have been proposed but such systems still face numerous challenges which impede their widespread adoption. In order to overcome many of these challenges, inspiration can be drawn from the signal processing and actuating techniques employed by the biological central nervous system to develop a bio-inspired control algorithm. In this study the front-end, signal processing techniques employed by biological sensory systems, and in particular the mammalian auditory system, are drawn upon in order to alleviate computations at the actuation node. This results in a simplistic control law that is a weighted combination of input information about the structure’s response such that F = WN, where F is the applied control force, W is a pre-determined weighting matrix, and N is a deconstructed representation of the structural response to the applied excitation. There is no empirical solution for deriving an optimal weighting matrix, W, and in this study numerous methods are explored in order to determine values for this matrix that produce the most effective control. These methods include particle swarm optimization, artificial neural networks, and optimal control theory. The various weighting matrices are integrated into the proposed bio-inspired control algorithm and applied in simulation to a five story benchmark structure. These methods are also compared to a traditional linear quadratic regulator (LQR) to gain insight into the overall performance of the bio-inspired control algorithm. Of the three training techniques, the particle swarm optimization technique offers the most effective control which is comparable in performance to the traditional LQR.