RESUMEN
Neural network training can be slow and energy-expensive due to the frequent transfer of weight data between digital memory and processing units. Neuromorphic systems can accelerate neural networks by performing multiply-accumulate operations in parallel using nonvolatile analog memory. However, executing the widely used backpropagation training algorithm in multilayer neural networks requires information-and therefore storage-of the partial derivatives of the weight values preventing suitable and scalable implementation in hardware. Here, we propose a hardware implementation of the backpropagation algorithm that progressively updates each layer using in situ stochastic gradient descent, avoiding this storage requirement. We experimentally demonstrate the in situ error calculation and the proposed progressive backpropagation method in a multilayer hardware-implemented neural network. We confirm identical learning characteristics and classification performance compared to conventional backpropagation in software. We show that our approach can be scaled to large and deep neural networks, enabling highly efficient training of advanced artificial intelligence computing systems.
RESUMEN
Biological systems interact directly with the environment and learn by receiving multimodal feedback via sensory stimuli that shape the formation of internal neuronal representations. Drawing inspiration from biological concepts such as exploration and sensory processing that eventually lead to behavioral conditioning, we present a robotic system handling objects through multimodal learning. A small-scale organic neuromorphic circuit locally integrates and adaptively processes multimodal sensory stimuli, enabling the robot to interact intelligently with its surroundings. The real-time handling of sensory stimuli via low-voltage organic neuromorphic devices with synaptic functionality forms multimodal associative connections that lead to behavioral conditioning, and thus the robot learns to avoid potentially dangerous objects. This work demonstrates that adaptive neuro-inspired circuitry with multifunctional organic materials, can accommodate locally efficient bio-inspired learning for advancing intelligent robotics.
Asunto(s)
Redes Neurales de la Computación , Robótica , Robótica/instrumentación , Robótica/métodos , Electrónica/instrumentación , Aprendizaje/fisiología , HumanosRESUMEN
Recent generative artificial intelligence (AI) has exerted a profound and far-reaching global impact across diverse fields and society. However, it comes at the cost of substantial energy and computational resource consumption. Neuromorphic computing endeavors to create highly efficient computing hardware that emulates biological neural networks and even mimics some human brain functions, and it is expected to play an essential role in the next-generation computing hardware. Memristors open up novel opportunities for neuromorphic computing due to their feasible ability to mimic neural functions. Innovation in memristors may lead to novel algorithms and contribute to conventionally challenging tasks like nondeterministic polynomial time (NP)-hard problem. To this end, we present a themed collection in Materials Horizons and Nanoscale Horizons, in which we publish the latest developments in memristive materials, device fabrication, characterization, and circuit design for neuromorphic systems.
RESUMEN
Signal communication mechanisms within the human body rely on the transmission and modulation of action potentials. Replicating the interdependent functions of receptors, neurons and synapses with organic artificial neurons and biohybrid synapses is an essential first step towards merging neuromorphic circuits and biological systems, crucial for computing at the biological interface. However, most organic neuromorphic systems are based on simple circuits which exhibit limited adaptability to both external and internal biological cues, and are restricted to emulate only specific the functions of an individual neuron/synapse. Here, we present a modular neuromorphic system which combines organic spiking neurons and biohybrid synapses to replicate a neural pathway. The spiking neuron mimics the sensory coding function of afferent neurons from light stimuli, while the neuromodulatory activity of interneurons is emulated by neurotransmitters-mediated biohybrid synapses. Combining these functions, we create a modular connection between multiple neurons to establish a pre-processing retinal pathway primitive.
Asunto(s)
Interneuronas , Neuronas , Humanos , Neuronas/fisiología , Potenciales de Acción/fisiología , Neuronas Aferentes , Sinapsis/fisiología , NeurotransmisoresRESUMEN
Accurate glucose prediction is vital for diabetes management. Artificial intelligence and artificial neural networks (ANNs) are showing promising results for reliable glucose predictions, offering timely warnings for glucose fluctuations. The translation of these software-based ANNs into dedicated computing hardware opens a route toward automated insulin delivery systems ultimately enhancing the quality of life for diabetic patients. ANNs are transforming this field, potentially leading to implantable smart prediction devices and ultimately to a fully artificial pancreas. However, this transition presents several challenges, including the need for specialized, compact, lightweight, and low-power hardware. Organic polymer-based electronics are a promising solution as they have the ability to implement the behavior of neural networks, operate at low voltage, and possess key attributes like flexibility, stretchability, and biocompatibility. Here, the study focuses on implementing software-based neural networks for glucose prediction into hardware systems. How to minimize network requirements, downscale the architecture, and integrate the neural network with electrochemical neuromorphic organic devices, meeting the strict demands of smart implants for in-body computation of glucose prediction is investigated.
Asunto(s)
Glucemia , Redes Neurales de la Computación , Humanos , Inteligencia Artificial , Glucemia/análisis , Programas InformáticosRESUMEN
Organic mixed ionic-electronic conductors (OMIECs) are central to bioelectronic applications such as biosensors, health-monitoring devices, and neural interfaces, and have facilitated efficient next-generation brain-inspired computing and biohybrid systems. Despite these examples, smart and adaptive circuits that can locally process and optimize biosignals have not yet been realized. Here, a tunable sensing circuit is shown that can locally modulate biologically relevant signals like electromyograms (EMGs) and electrocardiograms (ECGs), that is based on a complementary logic inverter combined with a neuromorphic memory element, and that is constructed from a single polymer mixed conductor. It is demonstrated that a small neuromorphic array based on this material effects high classification accuracy in heartbeat anomaly detection. This high-performance material allows for straightforward monolithic integration, which reduces fabrication complexity while also achieving high on/off ratios with excellent ambient p- and n-type stability in transistor performance. This material opens a route toward simple and straightforward fabrication and integration of more sophisticated adaptive circuits for future smart bioelectronics.
Asunto(s)
Técnicas Biosensibles , Transistores Electrónicos , Electrónica , Iones , PolímerosRESUMEN
In living organisms, sensory and motor processes are distributed, locally merged, and capable of forming dynamic sensorimotor associations. We introduce a simple and efficient organic neuromorphic circuit for local sensorimotor merging and processing on a robot that is placed in a maze. While the robot is exposed to external environmental stimuli, visuomotor associations are formed on the adaptable neuromorphic circuit. With this on-chip sensorimotor integration, the robot learns to follow a path to the exit of a maze, while being guided by visually indicated paths. The ease of processability of organic neuromorphic electronics and their unconventional form factors, in combination with education-purpose robotics, showcase a promising approach of an affordable, versatile, and readily accessible platform for exploring, designing, and evaluating behavioral intelligence through decentralized sensorimotor integration.
RESUMEN
Recent global events have distinctly demonstrated the need for fast diagnostic analysis of targets in a liquid sample. However, microfluidic lab-on-a-chip devices for point-of-care diagnostics can suffer from slow analysis due to poor mixing. Here, we experimentally explore the mixing effect within a microfluidic chamber, as obtained from superparamagnetic beads exposed to an out-of-plane (vertical) rotating magnetic field. Various magnetic protocols are explored, and the level of sample homogeneity is measured by determining the mixing efficiency index. In particular, we introduce a method to induce effective mixing in a microfluidic chamber by the actuation of the same beads to perform global swarming behavior, a collective motion of a large number of individual entities often seen in nature. The microparticle swarming induces high fluid velocities in initially stagnant fluids, and it can be externally controlled. The method is pilot-tested using a point-of-care test featuring a bioluminescent assay for the detection of antibodies. The mixing by the magnetic beads leads to increased assay kinetics, which indeed reduces the time to sensor readout substantially. Magnetic microparticle swarming is expected to be beneficial for a wide variety of point-of-care devices, where fast homogeneity of reagents does play a role.
Asunto(s)
Magnetismo , Microfluídica , Cinética , Dispositivos Laboratorio en un Chip , Campos MagnéticosRESUMEN
Brain-inspired computing paradigms have led to substantial advances in the automation of visual and linguistic tasks by emulating the distributed information processing of biological systems1. The similarity between artificial neural networks (ANNs) and biological systems has inspired ANN implementation in biomedical interfaces including prosthetics2 and brain-machine interfaces3. While promising, these implementations rely on software to run ANN algorithms. Ultimately, it is desirable to build hardware ANNs4,5 that can both directly interface with living tissue and adapt based on biofeedback6,7. The first essential step towards biologically integrated neuromorphic systems is to achieve synaptic conditioning based on biochemical signalling activity. Here, we directly couple an organic neuromorphic device with dopaminergic cells to constitute a biohybrid synapse with neurotransmitter-mediated synaptic plasticity. By mimicking the dopamine recycling machinery of the synaptic cleft, we demonstrate both long-term conditioning and recovery of the synaptic weight, paving the way towards combining artificial neuromorphic systems with biological neural networks.
Asunto(s)
Plasticidad Neuronal , Neurotransmisores/fisiología , Algoritmos , Animales , Redes Neurales de la Computación , Células PC12 , RatasRESUMEN
Organic electrochemical transistors (OECTs) show great promise for flexible, low-cost, and low-voltage sensors for aqueous solutions. The majority of OECT devices are made using the polymer blend poly(ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS), in which PEDOT is intrinsically doped due to inclusion of PSS. Because of this intrinsic doping, PEDOT:PSS OECTs generally operate in depletion mode, which results in a higher power consumption and limits stability. Here, a straightforward method to de-dope PEDOT:PSS using commercially available amine-based molecular de-dopants to achieve stable enhancement-mode OECTs is presented. The enhancement-mode OECTs show mobilities near that of pristine PEDOT:PSS (≈2 cm2 V-1 s-1 ) with stable operation over 1000 on/off cycles. The electron and proton exchange among PEDOT, PSS, and the molecular de-dopants are characterized to reveal the underlying chemical mechanism of the threshold voltage shift to negative voltages. Finally, the effect of the de-doping on the microstructure of the spin-cast PEDOT:PSS films is investigated.
RESUMEN
Microfluidic mixing becomes a necessity when thorough sample homogenization is required in small volumes of fluid, such as in lab-on-a-chip devices. For example, efficient mixing is extraordinarily challenging in capillary-filling microfluidic devices and in microchambers with stagnant fluids. To address this issue, specifically designed geometrical features can enhance the effect of diffusion and provide efficient mixing by inducing chaotic fluid flow. This scheme is known as "passive" mixing. In addition, when rapid and global mixing is essential, "active" mixing can be applied by exploiting an external source. In particular, magnetic mixing (where a magnetic field acts to stimulate mixing) shows great potential for high mixing efficiency. This method generally involves magnetic beads and external (or integrated) magnets for the creation of chaotic motion in the device. However, there is still plenty of room for exploiting the potential of magnetic beads for mixing applications. Therefore, this review article focuses on the advantages of magnetic bead mixing along with recommendations on improving mixing in low Reynolds number flows (Re ≤ 1) and in stagnant fluids.
RESUMEN
Poly(3,4-ethylenedioxythiophene) blended with polystyrenesulfonate and poly(styrenesulfonic acid), PEDOT:PSS, has found widespread use in organic electronics. Although PEDOT:PSS is commonly used in its doped electrically conducting state, the ability to efficiently convert PEDOT:PSS to its undoped nonconducting state is of interest for a wide variety of applications ranging from biosensors to organic neuromorphic devices. Exposure to aliphatic monoamines, acting as an electron donor and Brønsted-Lowry base, has been reported to be partly successful, but monoamines are unable to fully dedope PEDOT:PSS. Remarkably, some-but not all-polyamines can dedope PEDOT:PSS very efficiently to very low conductivity levels, but the exact chemical mechanism involved is not understood. Here, we study the dedoping efficacy of 21 different aliphatic amines. We identify the presence of two or more primary amines, which can participate in an intramolecular reaction, as the key structural motif that endows polyamines with high PEDOT:PSS dedoping strength. A multistep reaction mechanism, involving sequential electron transfer and deprotonation steps, is proposed that consistently explains the experimental results. Finally, we provide a simple method to convert the commonly used aqueous PEDOT:PSS dispersion into a precursor formulation that forms fully dedoped PEDOT:PSS films after spin coating and subsequent thermal annealing.
RESUMEN
A variety of optical applications rely on the absorption and reemission of light. The quantum yield of this process often plays an essential role. When the quantum yield deviates from unity by significantly less than 1%, applications such as luminescent concentrators and optical refrigerators become possible. To evaluate such high performance, we develop a measurement technique for luminescence efficiency with sufficient accuracy below one part per thousand. Photothermal threshold quantum yield is based on the quantization of light to minimize overall measurement uncertainty. This technique is used to guide a procedure capable of making ensembles of near-unity emitting cadmium selenide/cadmium sulfide (CdSe/CdS) core-shell quantum dots. We obtain a photothermal threshold quantum yield luminescence efficiency of 99.6 ± 0.2%, indicating nearly complete suppression of nonradiative decay channels.
RESUMEN
Interfacing soft materials with biological systems holds considerable promise for both biosensors and recording live cells. However, the interface between cells and organic substrates is not well studied, despite its crucial role in the effectiveness of the device. Furthermore, well-known cell adhesion enhancers, such as microgrooves, have not been implemented on these surfaces. Here, we present a nanoscale characterization of the cell-substrate interface for 3D laser-patterned organic electrodes by combining electrochemical impedance spectroscopy (EIS) and scanning electron microscopy/focused ion beam (SEM/FIB). We demonstrate that introducing 3D micropatterned grooves on organic surfaces enhances the cell adhesion of electrogenic cells.
Asunto(s)
Rayos Láser , Compuestos Bicíclicos Heterocíclicos con Puentes , Adhesión Celular , Electrodos , Microscopía Electrónica de Rastreo , Polímeros , Factores de TiempoRESUMEN
The interface between cells and nonbiological surfaces regulates cell attachment, chronic tissue responses, and ultimately the success of medical implants or biosensors. Clinical and laboratory studies show that topological features of the surface profoundly influence cellular responses; for example, titanium surfaces with nano- and microtopographical structures enhance osteoblast attachment and host-implant integration as compared to a smooth surface. To understand how cells and tissues respond to different topographical features, it is of critical importance to directly visualize the cell-material interface at the relevant nanometer length scale. Here, we present a method for in situ examination of the cell-to-material interface at any desired location, based on focused ion beam milling and scanning electron microscopy imaging to resolve the cell membrane-to-material interface with 10 nm resolution. By examining how cell membranes interact with topographical features such as nanoscale protrusions or invaginations, we discovered that the cell membrane readily deforms inward and wraps around protruding structures, but hardly deforms outward to contour invaginating structures. This asymmetric membrane response (inward vs outward deformation) causes the cleft width between the cell membrane and the nanostructure surface to vary by more than an order of magnitude. Our results suggest that surface topology is a crucial consideration for the development of medical implants or biosensors whose performances are strongly influenced by the cell-to-material interface. We anticipate that the method can be used to explore the direct interaction of cells/tissue with medical devices such as metal implants in the future.
RESUMEN
The brain is capable of massively parallel information processing while consuming only â¼1-100 fJ per synaptic event. Inspired by the efficiency of the brain, CMOS-based neural architectures and memristors are being developed for pattern recognition and machine learning. However, the volatility, design complexity and high supply voltages for CMOS architectures, and the stochastic and energy-costly switching of memristors complicate the path to achieve the interconnectivity, information density, and energy efficiency of the brain using either approach. Here we describe an electrochemical neuromorphic organic device (ENODe) operating with a fundamentally different mechanism from existing memristors. ENODe switches at low voltage and energy (<10 pJ for 103 µm2 devices), displays >500 distinct, non-volatile conductance states within a â¼1 V range, and achieves high classification accuracy when implemented in neural network simulations. Plastic ENODes are also fabricated on flexible substrates enabling the integration of neuromorphic functionality in stretchable electronic systems. Mechanical flexibility makes ENODes compatible with three-dimensional architectures, opening a path towards extreme interconnectivity comparable to the human brain.
Asunto(s)
Encéfalo , Computadores Moleculares , Técnicas Electroquímicas , Red Nerviosa , HumanosRESUMEN
Laser assisted catalytic chemical vapor deposition has recently emerged as an attractive method for locally growing carbon nanotubes (CNTs) in a cold wall reactor. So far, reported laser assisted CNT growth has been carried out without in situ process monitoring. This has made it difficult to control the growth process and limits the applicability of the method. Using a set of photodetectors with different spectral responses, we show that one can identify characteristic regimes of laser assisted CNT growth. More specifically, key process steps like catalyst activation, growth of CNTs, and amorphous carbon deposition are identified. Furthermore, the method allows optimization of growth conditions with respect to the quality of the growth products.