International Business Machines Corporation

Siegfried Friedrich Karg

Heike E Riel

Walter Heinrich Riess

Gerhard Ingmar Meijer

Bernd W Gotsmann

Heinz Schmid

Johannes Georg Bednorz

Mikael T Bjoerk

Joachim Knoch

Tilman A Beierlein

Bruno Michel

Thomas J Brunschwiler

Lukas Czornomaz

Cezar Bogdan Zota

Roland Germann

Clarissa Convertino

David J Gundlach

Siegfried Johannes Barth

Elisabetta Corti

Chung H Lam

Paul Michael Solomon

Alejandro Gabriel Schrott

Eric Andrew Joseph

Peter Mueller

Emanuel Loertscher

Ute Drechsler

Hajime Nakamura

Fabian Menges

Ingmar G Meijer

Reto R Schlittler

Paul F Seidler

Volker Schmidt

Giorgio Signorello

Santos F Alvarado

Roland W Widmer

Lieike Riel

Beat Ruhstaller

Gerhard J Meijer

Constance Rost-Bietsch

Gerhard Ingmar Meijier

Brian K Crone

Roland W Germann

The network comprises at least one network layer in which a plurality of electronic oscillators, interconnected via programmable coupling elements storing respective network weights, generate oscillatory signals at time delays dependent on the input signal to propagate the input signal from an input to an output of that layer. The network is adapted to provide a network output signal dependent substantially linearly on phase of oscillatory signals in the last layer of the network. The method includes calculating a network error dependent on the output signal and a desired output for the training sample, and calculating updates for respective network weights by backpropagation of the error such that weight-updates for a network layer are dependent on a vector of time delays at the input to that layer and the calculated error at the output of that layer.

Training of oscillatory neural networks

A device includes at least one tunable resistive element. Each tunable resistive element comprises a first terminal, a second terminal, and a dielectric layer arranged between the first and second terminals. The device is configured to apply at least one electrical set pulse to the resistive elements to form a conductive filament comprising a plurality of oxygen vacancies in the dielectric layer. The device is configured to apply at least one electrical reset pulse to displace a subset of the oxygen vacancies of the conductive filament. The at least one electrical reset pulse comprises a first part, which is adapted to increase the temperature of the conductive filament and increase the mobility of the oxygen vacancies of the conductive filament, and a second part, which is configured to displace the subset of the oxygen vacancies of the conductive filament.

Device comprising tunable resistive elements

The invention relates to a capacitorless DRAM cell, the cell comprising a heterostructure, a gate structure adjoining the heterostructure in a first direction, a drain structure adjoining the heterostructure in a second direction perpendicular to the first direction, and a source structure adjoining the heterostructure in the direction opposite the second direction, the heterostructure comprising one or more semiconducting channel layers and one or more electrically insulating barrier layers, the channel layers and the barrier layers being alternatingly stacked in the first direction.

Capacitorless DRAM cell

An electronic circuit for enabling an efficient use of an oscillating neural network for feature recognition may be provided. The electronic circuit comprises a network of coupled voltage-controlled oscillators, wherein each of the voltage-controlled oscillators is adapted for receiving an edge input signal which is phase-shifted by a fraction of a period length of the voltage-controlled oscillators according to a signal strength of an analog input signal allotted to a respective one of the voltage-controlled oscillators, and an active output circuit. The active output circuit includes input terminals connected to selected ones of the voltage-controlled oscillators, an adder portion for adding input signals present at the input terminals, and a non-linear amplifier, an which input line is of the non-linear amplifier being connected to an output line of the adder portion, thereby an efficient use of an oscillating neural network.

Feature recognition with oscillating neural network

A computing network comprising a first layer comprising a first set of oscillators and a second layer comprising a second set of oscillators is provided. The computing network further comprises a plurality of adjustable coupling elements between the oscillators of the first set and a plurality of nonlinear elements. The nonlinear elements are configured to couple output signals of the first layer to the second layer. The computing network further comprises an encoding unit configured to receive input signals, convert the input signals into phase-encoded output signals, and provide the phase-encoded output signals to the first layer.

Multi-layer oscillating network

Capacitorless dram cell

An optical spectrometer contains a photodiode and a straining mechanism for imposing adjustable strain on the photodiode. The spectrometer includes a measurement apparatus for measuring variation of photocurrent with strain at different values of the adjustable strain imposed by the straining mechanism. Adjusting the strain allows adjustment of the band gap Eof the photosensitive region of the photodiode, and this determines the cut-off energy for absorption of photons. Measuring variation of photocurrent with strain at different values of the adjustable strain imposed by the straining mechanism allows study of photons within a desired energy range of the band gap energy corresponding to each strain value.

Optical spectrometer

A synapse for a neuromorphic network is provided. The synapse includes a time-delay portion having a first end and a second end, a first actuator located at the first end and in operational contact with the time-delay portion, and a second actuator located at the second end and in operational contact with the time-delay portion. The time-delay portion is formed from a phase change material wherein a change in the material of the time-delay portion alters a propagation time of a signal transmitted from the first actuator to the second actuator.

Phase-change material time-delay element for neuromorphic networks

A thermoelectric element includes a body formed of a single thermoelectric material and extending in a first direction along which a thermal gradient is established in thermoelectric operation, wherein the body has at least first and second adjacent sections in the first direction; at least one of the sections is subject to stress which is applied to that section substantially all around a central axis of the body in the first direction; and the arrangement is such that the stress results in different strain in the first and second sections producing an energy barrier in the body to enhance thermoelectric operation.

Thermoelectric elements

A method of forming nanowire devices. The method includes forming a stressor layer circumferentially surrounding a semiconductor nanowire. The method is performed such that, due to the stressor layer, the nanowire is subjected to at least one of radial and longitudinal strain to enhance carrier mobility in the nanowire. Radial and longitudinal strain components can be used separately or together and can each be made tensile or compressive, allowing formulation of desired strain characteristics for enhanced conductivity in the nanowire of a given device.

Nanowire devices

A solar cell includes a substrate layer and a plurality of nanowires grown outwardly from the substrate layer, at least two of the nanowires including a plurality of sub-cells. The solar cell also includes one or more light guiding layers formed of a transparent, light scattering material and filling the area between the plurality of nanowires.

Nanowire multijunction solar cell

A multi-terminal programmable element. The programmable element includes a source electrode and a drain electrode on a base. The programmable element includes reference voltage contact that is not in contact with the source or drain electrode. The base includes a transition-metal oxide with oxygen vacancies for drifting under an applied electric field. Further, materials of the source electrode and the base are selected such that an interface of a source and/or drain electrode material and the transition metal oxide base material forms an energy barrier for electron injection from the electrode into the base material. The energy barrier has a height that depends on an oxygen vacancy concentration of the base material. Four non-volatile states are programmable into the programmable element.

Programmable element, and memory device or logic circuit

A complementary logic element including first and second transistor elements. The first and second gate electrodes of the two transistor elements are electrically parallel to form a common gate. Both the coupling layers of the first and the second transistor element include a resistance switching material, a conductivity of which may be altered by causing an ion concentration to alter if an electrical voltage signal of an appropriate polarity is applied. The first and second transistor elements also include an ion conductor layer that is capable of accepting ions from the coupling layer and of releasing ions into the coupling layer. The coupling layers and ion conductor layers are such that the application of an electrical signal of a given polarity to the gate enhances the electrical conductivity of the first coupling layer and diminishes the electrical conductivity of the second, or vice versa.

Logic element, and integrated circuit or field programmable gate array

Growing community of inventors

Adliswil, Switzerland

Siegfried Friedrich Karg