Bio-inspired Information Processing Research Articles

Bio-inspired artificial synaptic transistors: evolution from innovative basic units to system integration.

The investigation of transistor-based artificial synapses in bioinspired information processing is undergoing booming exploration, and is the stable building block for brain-like computing. Given that the storage and computing separation architecture of von Neumann construction is not conducive to the current explosive information processing, it is critical to accelerate the connection between hardware systems and software simulations of intelligent synapses. So far, various works based on a transistor-based synaptic system successfully simulated functions similar to biological nerves in the human brain. However, the influence of the semiconductor and the device structural design on synaptic properties is still poorly linked. This review concretely emphasizes the recent advances in the novel structure design of semiconductor materials and devices used in synaptic transistors, not only from a single multifunction synaptic device but also to system application with various connected routes and related working mechanisms. Finally, crises and opportunities in transistor-based synaptic interconnection are discussed and predicted.

An Aqueous Analog MAC Machine.

Using ions in aqueous milieu for signal processing, like in biological circuits, may potentially lead to a bioinspired information processing platform. Studies, however, have focused on individual ionic diodes and transistors rather than circuits comprising many such devices. Here a 16×16 array of new ionic transistors is developed in an aqueous quinone solution. Each transistor features a concentric ring electrode pair with a disk electrode at the center. The electrochemistry of these electrodes in the solution provides the basis for the transistor operation. The ring pair electrochemically tunes the local electrolytic concentration to modulate the disk's Faradaic reaction rate. Thus, the disk current as a Faradaic reaction to the disk voltage is gated by the ring pair. The 16×16 array of these transistors performs analog multiply-accumulate (MAC) operations, a computing modality hotly pursued for low-power artificial neural networks. This exploits the transistor's operating regime where the disk current is a multiplication of the disk voltage and a weight parameter tuned by the ring pair gating. Such disk currents from multiple transistors are summated in a global reference electrode to complete a MAC task. This ionic circuit demonstrating analog computing is a step toward sophisticated aqueous ionics.

Decoding complex state space trajectories for neural computing.

In biological neural circuits as well as in bio-inspired information processing systems, trajectories in high-dimensional state-space encode the solutions to computational tasks performed by complex dynamical systems. Due to the high state-space dimensionality and the number of possible encoding trajectories rapidly growing with input signal dimension, decoding these trajectories constitutes a major challenge on its own, in particular, as exponentially growing (space or time) requirements for decoding would render the original computational paradigm inefficient. Here, we suggest an approach to overcome this problem. We propose an efficient decoding scheme for trajectories emerging in spiking neural circuits that exhibit linear scaling with input signal dimensionality. We focus on the dynamics near a sequence of unstable saddle states that naturally emerge in a range of physical systems and provide a novel paradigm for analog computing, for instance, in the form of heteroclinic computing. Identifying simple measures of coordinated activity (synchrony) that are commonly applicable to all trajectories representing the same percept, we design robust readouts whose sizes and time requirements increase only linearly with the system size. These results move the conceptual boundary so far hindering the implementation of heteroclinic computing in hardware and may also catalyze efficient decoding strategies in spiking neural networks in general.

Organic Synaptic Transistors: The Evolutionary Path from Memory Cells to the Application of Artificial Neural Networks

AbstractThe progress of neural synaptic devices is experiencing an era of explosive growth. Given that the traditional storage system has yet to overcome the von Neumann bottleneck, it is critical to develop hardware with bioinspired information processing functions and lower power consumption. Transistors based on 2D materials, metal oxides, and organic materials have been adopted to mimic the synapse of a human brain, due to their high plasticity, parallel computing, integrated storage, and system information processing. Among these materials used to build transistors, organic semiconductors are considered to be the most promising candidate for neural synaptic devices and bio‐electronics, owing to their easy processing, mechanical flexibility, low cost, good bio‐compatibility, and ductility. This review focuses on the recent advances in organic synaptic devices with various structures, materials, and working mechanisms. The applications of artificial neural networks that integrate multiple organic synaptic transistors are also concretely discussed. Finally, the challenges that organic synaptic devices currently face are discussed and future developments are forecast.

A 4-fJ/Spike Artificial Neuron in 65 nm CMOS Technology.

As Moore's law reaches its end, traditional computing technology based on the Von Neumann architecture is facing fundamental limits. Among them is poor energy efficiency. This situation motivates the investigation of different processing information paradigms, such as the use of spiking neural networks (SNNs), which also introduce cognitive characteristics. As applications at very high scale are addressed, the energy dissipation needs to be minimized. This effort starts from the neuron cell. In this context, this paper presents the design of an original artificial neuron, in standard 65 nm CMOS technology with optimized energy efficiency. The neuron circuit response is designed as an approximation of the Morris-Lecar theoretical model. In order to implement the non-linear gating variables, which control the ionic channel currents, transistors operating in deep subthreshold are employed. Two different circuit variants describing the neuron model equations have been developed. The first one features spike characteristics, which correlate well with a biological neuron model. The second one is a simplification of the first, designed to exhibit higher spiking frequencies, targeting large scale bio-inspired information processing applications. The most important feature of the fabricated circuits is the energy efficiency of a few femtojoules per spike, which improves prior state-of-the-art by two to three orders of magnitude. This performance is achieved by minimizing two key parameters: the supply voltage and the related membrane capacitance. Meanwhile, the obtained standby power at a resting output does not exceed tens of picowatts. The two variants were sized to 200 and 35 μm2 with the latter reaching a spiking output frequency of 26 kHz. This performance level could address various contexts, such as highly integrated neuro-processors for robotics, neuroscience or medical applications.

Implementing Signature Neural Networks with Spiking Neurons.

Spiking Neural Networks constitute the most promising approach to develop realistic Artificial Neural Networks (ANNs). Unlike traditional firing rate-based paradigms, information coding in spiking models is based on the precise timing of individual spikes. It has been demonstrated that spiking ANNs can be successfully and efficiently applied to multiple realistic problems solvable with traditional strategies (e.g., data classification or pattern recognition). In recent years, major breakthroughs in neuroscience research have discovered new relevant computational principles in different living neural systems. Could ANNs benefit from some of these recent findings providing novel elements of inspiration? This is an intriguing question for the research community and the development of spiking ANNs including novel bio-inspired information coding and processing strategies is gaining attention. From this perspective, in this work, we adapt the core concepts of the recently proposed Signature Neural Network paradigm—i.e., neural signatures to identify each unit in the network, local information contextualization during the processing, and multicoding strategies for information propagation regarding the origin and the content of the data—to be employed in a spiking neural network. To the best of our knowledge, none of these mechanisms have been used yet in the context of ANNs of spiking neurons. This paper provides a proof-of-concept for their applicability in such networks. Computer simulations show that a simple network model like the discussed here exhibits complex self-organizing properties. The combination of multiple simultaneous encoding schemes allows the network to generate coexisting spatio-temporal patterns of activity encoding information in different spatio-temporal spaces. As a function of the network and/or intra-unit parameters shaping the corresponding encoding modality, different forms of competition among the evoked patterns can emerge even in the absence of inhibitory connections. These parameters also modulate the memory capabilities of the network. The dynamical modes observed in the different informational dimensions in a given moment are independent and they only depend on the parameters shaping the information processing in this dimension. In view of these results, we argue that plasticity mechanisms inside individual cells and multicoding strategies can provide additional computational properties to spiking neural networks, which could enhance their capacity and performance in a wide variety of real-world tasks.

Can bio-inspired information processing steps be realized as synthetic biochemical processes?

We consider possible designs and experimental realiza-tions in synthesized rather than naturally occurring bio-chemical systems of a selection of basic bio-inspired information processing steps. These include feed-forward loops, which have been identified as the most common information processing motifs in many natural pathways in cellular functioning, and memory-involving processes, specifically, associative memory. Such systems should not be designed to literally mimic nature. Rather, we can be guided by nature's mechanisms for experimenting with new information/signal processing steps which are based on coupled biochemical reactions, but are vastly simpler than natural processes, and which will provide tools for the long-term goal of understanding and harnessing nature's information processing paradigm. Our biochemical processes of choice are enzymatic cascades because of their compatibility with physiological processes in vivo and with electronics (e.g., electrodes) in vitro allowing for networking and interfacing of enzyme-catalyzed processes with other chemical and biochemical reactions. In addition to designing and realizing feed-forward loops and other processes, one has to develop approaches to probe their response to external control of the time-dependence of the input(s), by measuring the resulting time-dependence of the output. The goal will be to demonstrate the expected features, for example, the delayed response and stabilizing effect of the feed-forward loops.

Complex photonics: Dynamics and applications of delay-coupled semiconductors lasers

Complex phenomena in photonics, in particular, dynamical properties of semiconductor lasers due to delayed coupling, are reviewed. Although considered a nuisance for a long time, these phenomena now open interesting perspectives. Semiconductor laser systems represent excellent test beds for the study of nonlinear delay-coupled systems, which are of fundamental relevance in various areas. At the same time delay-coupled lasers provide opportunities for photonic applications. In this review an introduction into the properties of single and two delay-coupled lasers is followed by an extension to network motifs and small networks. A particular emphasis is put on emerging complex behavior, deterministic chaos, synchronization phenomena, and application of these properties that range from encrypted communication and fast random bit sequence generators to bioinspired information processing.

Memcapacitive properties of poly(3,4-ethylenedioxythiophene) modified electrodes

We study the capacitive properties of poly(3,4-ethylenedioxythiophene) (PEDOT) thin layer that shows memory effects. The capacitance vs. charge plot indicates the history-dependence of the electrochemical responses, for instance capacitance with memory or memcapacitance. The variation of the charge vs. potential exhibits a hysteresis loop that is related to the non-equilibrium properties of the redox switching. The calculation of the energy that can be added to and removed from the memcapacitance system during cycling shows a reactive behavior. This property is due to the electromechanical behaviors of PEDOT film that can store and transform electrochemical energy. Indeed, the charging–discharging processes during the redox switching induce a volume and conformational change of the polymer that in turn influences its electrochemical responses. In the context of bio-inspired information processing systems or bio-inspired circuits, a 3-terminal device exhibiting non-linearity and history-dependence responses is of interest for developing organic memory-circuit elements such as an electrochemical memcapacitance.

Modeling of Synchronization in Bushy Cells of Cochlear Nucleus

Event Abstract Back to Event Modeling of Synchronization in Bushy Cells of Cochlear Nucleus Marek Rudnicki1 and Werner Hemmert1* 1 Technische Universität München, Bio-Inspired Information Processing, Germany In the inner ear sounds are converted into action potentials and propagated to the central nervous system. The cochlear nucleus in the auditory brainstem is the first station that receives synaptic inputs from auditory nerve fibers (ANF) and consists of several neuron types. Here we focus on globular bushy cells (GBC) that are part of the sound localization pathway. One of the main properties of GBC is their excellent temporal firing precision in response to sounds.Our model of GBCs is a point neuron with Hodgkin-Huxley type ion channels (HPAC, Kht, Klt). It receives several excitatory inputs from an inner ear model simulating responses of ANFs. ANF activity drives endbulb of Held synapses located directly on the GBC soma. We modeled two types of synapses: with short-term depression as measured in many in-vitro studies (Yang and Xu-Friedman 2009) and without depression as reported by recent in-vivo studies (Borst 2010).The model is able to reproduce standard experiments involving pure tone stimulation. In particular the PSTH displays proper levels of spontaneous and driven rates as well as characteristic primary-like with notch shapes. Additionally, stimulation with low frequency tones shows improvement in synchronization that can be characterized as high-sync (SI > 0.9) according to Joris et al. (1994). Interestingly high entrainment levels where achieved only by the model without synaptic depression.In summary, our model is able to reproduce the key features of GBC responses. Our results suggest that depression in-vivo is much lower compared to in-vitro conditions. In the future, the model will allow us to study the response of GBCs to complex natural stimuli like speech. Acknowledgements This work was funded within BCCN Munich by the German Federal Ministry of Education and Research (reference number 01GQ0441). Keywords: computational neuroscience Conference: Bernstein Conference on Computational Neuroscience, Berlin, Germany, 27 Sep - 1 Oct, 2010. Presentation Type: Presentation Topic: Bernstein Conference on Computational Neuroscience Citation: Rudnicki M and Hemmert W (2010). Modeling of Synchronization in Bushy Cells of Cochlear Nucleus. Front. Comput. Neurosci. Conference Abstract: Bernstein Conference on Computational Neuroscience. doi: 10.3389/conf.fncom.2010.51.00015 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 22 Sep 2010; Published Online: 23 Sep 2010. * Correspondence: Prof. Werner Hemmert, Technische Universität München, Bio-Inspired Information Processing, München, Germany, werner.hemmert@tum.de Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Marek Rudnicki Werner Hemmert Google Marek Rudnicki Werner Hemmert Google Scholar Marek Rudnicki Werner Hemmert PubMed Marek Rudnicki Werner Hemmert Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.

CMOS-Based Bio/Chemosensor and Bioelectronic Microsystems

Microfabrication techniques and, in particular, complementary-metal-oxide-semiconductor (CMOS) technology have been used to devise chemo/biosensors [1-3] as well as bioelectronic microsystems [4-7]. Examples of micromachined bio- or chemosensors, such as cantilevers or micoelectrode arrays, will be shown, and the electrical interfacing of CMOS microelectronics with biological entities or electrogenic cells, i.e., cells that react upon electrical stimulation and, in turn, produce electrical signals (heart cells or neurons) are detailed. CMOS-based, fully integrated microelectrode arrays for bidirectional communication (stimulation and recording) with electrogenic cells are presented. These devices are capable of monitoring relevant electrophysiological responses of cells to electrical stimuli or to pharmacological agents with prospective applications in the field of bio-inspired information processing or pharmascreening.

Optimization of an organic memristor as an adaptive memory element

The combination of memory and signal handling characteristics of a memristor makes it a promising candidate for adaptive bioinspired information processing systems. This poses stringent requirements on the basic device, such as stability and reproducibility over a large number of training/learning cycles, and a large anisotropy in the fundamental control material parameter, in our case the electrical conductivity. In this work we report results on the improved performance of electrochemically controlled polymeric memristors, where optimization of a conducting polymer (polyaniline) in the active channel and better environmental control of fabrication methods led to a large increase both in the absolute values of the conductivity in the partially oxydized state of polyaniline and of the on-off conductivity ratio. These improvements are crucial for the application of the organic memristor to adaptive complex signal handling networks.

Hybrid intelligent and adaptive sensor systems with improved signal integrity based on dynamically reconfigurable matched universal sensor electronics

Hybrid intelligent sensor systems and networks are composed of modules of tightly co-operating software and hardware components. Bio-inspired information processing is embodied in algorithms as well as dedicated electronics for intelligent processing