Neuromorphic computers could overcome efficiency bottlenecks inherent to conventional computing through parallel programming and readout of artificial neural network weights in a crossbar memory array. However, selective linear weight updates <10-nanoampere read currents are required for learning that surpasses efficiency. We introduce an ionic floating-gate array based on polymer redox transistor connected conductive-bridge (CBM). Selective is executed by overcoming the bridging threshold...

Nonvolatile redox transistors (NVRTs) based upon Li-ion battery materials are demonstrated as memory elements for neuromorphic computer architectures with multi-level analog states, "write" linearity, low-voltage switching, and low power dissipation. Simulations of backpropagation using the device properties reach ideal classification accuracy. Physics-based simulations predict energy costs per operation <10 aJ when scaled to 200 nm × nm.

Resistive memories enable dramatic energy reductions for neural algorithms. We propose a general purpose architecture that can accelerate many different algorithms and determine the device properties will be needed to run backpropagation on architecture. To maintain high accuracy, read noise standard deviation should less than 5% of weight range. The write 0.4% range up 300% characteristic update (for datasets tested). Asymmetric nonlinearities in change conductance vs pulse cause decay...

Microcontact printing (μCP) is a new method of molecularly patterning surfaces on micrometer scale. In this paper, we present the extension microcontact to producing patterned layers proteins solid substrates. μCP avoids use strong acids and bases necessary in photolithographic patterning, allowing its for other biological layers. We also describe methods thin stamp that allow isolated features previously unattainable by printing. A solution polylysine borate-buffered saline was printed onto...

We describe a method for producing high-resolution chemical patterns on surfaces to control the attachment and growth of cultured neurons. Microcontact printing has been extended allow /spl mu/m-scale protein lines aligned an underlying pattern planar microelectrodes. Poly-L-lysine (PL) have printed electrode array electrical studies neural networks. Rat hippocampal neurons showed degree selectivity PL produced neurites that faithfully grew onto recording sites.

We have developed a method for performing light-sheet microscopy with single high numerical aperture lens by integrating reflective side walls into microfluidic chip. These 45° generate illumination reflecting vertical the focal plane of objective. Light-sheet cells loaded in channels increases image quality diffraction limited imaging via reduction out-of-focus background light. Single molecule super-resolution is also improved decreased resulting better localization precision and...

The exponential increase in data over the last decade presents a significant challenge to analytics efforts that seek process and interpret such for various applications. Neural-inspired computing approaches are being developed order leverage computational properties of analog, low-power processing observed biological systems. Analog resistive memory crossbars can perform parallel read or vector-matrix multiplication as well write rank-1 update with high efficiency. For an N × crossbar,...

The brain has effectively proven a powerful inspiration for the development of computing architectures in which processing is tightly integrated with memory, communication event-driven, and analog computation can be performed at scale. These neuromorphic systems increasingly show an ability to improve efficiency speed scientific artificial intelligence applications. Herein, it proposed that brain's ubiquitous stochasticity represents additional source expanding reach probabilistic To date,...

Neuronal cell networks have been reconstructed on planar microelectrode arrays (MEAs) from dissociated hippocampal pyramidal neurons. Microcontact printing (microCP) and a photoresist-liftoff method were used to selectively localize poly-L-lysine (PLL) the surface of MEAs. Haptotaxis led organization neurons into localized adjacent microelectrodes. Various grids PLL with 2-25-microm-wide lines spaced by 50-200 microm 15-25-microm nodes at intersection points guide body attachment neurite...

Neural networks are an increasingly attractive algorithm for natural language processing and pattern recognition. Deep with >50M parameters made possible by modern GPU clusters operating at <50 pJ per op more recently, production accelerators capable of <5pJ operation the board level. However, slowing CMOS scaling, new paradigms will be required to achieve next several orders magnitude in performance watt gains. Using analog resistive memory (ReRAM) crossbar perform key matrix operations...

Neural machine learning methods, such as deep neural networks (DNN), have achieved remarkable success in a number of complex data processing tasks. These methods arguably had their strongest impact on tasks image and audio - domains which humans long held clear advantages over conventional algorithms. In contrast to biological systems, are capable continuously, artificial limited ability for incorporating new information an already trained network. As result, continuous potentially highly...

We present resistive switching model for TaOx memristors, which demonstrates that the radius of a tantalum rich conducting filament is state variable controlling resistance. The tracks flux individual oxygen ions and permits derivation solving dynamical static equations. Model predictions ON/OFF were tested experimentally with devices shown to be in close quantitative agreement, including observed transition from linear non-linear conduction between RON ROFF. This work presents dynamics...

The steady-state solution of filamentary memristive switching may be derived directly from the heat equation, modelling vertical and radial flow. This is shown to provide a continuous accurate description evolution filament radius, composition, flow, temperature during switching, apply large range materials experimental time-scales. As service our authors readers, this journal provides supporting information supplied by authors. Such are peer reviewed re-organized for online delivery, but...

Analog resistive memories promise to reduce the energy of neural networks by orders magnitude. However, write variability and nonlinearity current devices prevent from training high accuracy. We present a novel periodic carry method that uses positional number system overcome this while maintaining benefit parallel analog matrix operations. demonstrate how noisy, nonlinear TaO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">x</sub> could only...

Recent cyber security events have demonstrated the need for algorithms that adapt to rapidly evolving threat landscape of complex network systems. In particular, human analysts often fail identify data exfiltration when it is encrypted or disguised as innocuous data. Signature-based approaches identifying types are easily fooled and can only investigate a small fraction events. However, neural networks learn subtle patterns in suitably chosen input space. To this end, we developed signal...

Nanobiotechnology is a field that utilizes the techniques of nano- and microfabrication to study biosystems or use biological material principles build new devices. As an example we discuss development nanofabricated electrochemical detector array reveals spatio-temporal dynamics exocytosis in single chromaffin cells. In quantal release event vesicle fuses with plasma membrane releasing its contents through fusion pore. The time-resolved amperometric currents measured by individual...

File fragment classification is an important step in the task of file carving digital forensics. In carving, files must be reconstructed based on their content as a result fragmented storage disk or memory. Existing methods for fragments typically use hand-engineered features, such byte histograms entropy measures. this paper, we propose approach using sparse coding that enables automated feature extraction. Sparse coding, dictionary learning, unsupervised learning algorithm, and capable...

We have developed a pulsed optically pumped magnetometer (OPM) array for detecting magnetic field maps originated from an arbitrary current distribution. The presented source imaging (MSI) system features 24-OPM channels has data rate of 500 S/s, sensitivity <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.8~\mathrm {pT/}\sqrt {\mathrm {Hz}} $ </tex-math></inline-formula> , and dynamic range 72 dB....

Abstract Beyond-von Neumann computing approaches are necessary to sustain the growth of microelectronics and increasing appetite for artificial intelligence/machine learning algorithms. Neuromorphic is an emerging paradigm that takes inspiration from brain provide a path forward improve computational efficiency density next-generation architectures.&amp;#xD;&amp;#xD; In nature, we observe brains performing complex computations with much smaller energy footprint than conventional approaches....