This method for training a binarized neural network, also called BNN, including neurons, with a binary weight for each connection between two neurons, is implemented by an electronic circuit and comprises: a forward pass including calculating an output vector by applying the BNN on an input vector; a backward pass including computing an error vector from the calculated output vector, and calculating a new value of the input vector by applying the BNN on the error vector; a weight update including computing a product by multiplying an element of the error vector with one of the new value of the input vector, modifying a latent variable depending on the product; and updating the weight with the latent variable;

each weight being encoded using a primary memory component;

each latent variable being encoded using a secondary memory component having a characteristic subject to a time drift.

A memory includes read circuitry for reading values stored in memory cells. The read circuitry includes flipped voltage followers for providing bias voltages to nodes of current paths coupled to sense amplifiers during memory read operations.

A storage circuit (11) includes memory cells (MCij), each of which includes an MTJ element, and reference cells (RCi), each of which includes a series circuit of an MTJ element set to a low-resistance state and a linear resistor (FR). A RW circuit (23j) that includes a sense amplifier is provided in each column of a memory cell array (21), and compares a data voltage on a corresponding bit line (BLj) with a reference voltage. The sense amplifier includes a pair of PMOS transistors to which the data voltage and the reference voltage are applied, a CMOS sense latch that is connected to a current path of the PMOS transistors.

A memory circuit configured to perform multiply-accumulate (MAC) operations for performance of an artificial neural network includes a series of synapse cells arranged in a cross-bar array. Each cell includes a memory transistor connected in series with a memristor. The memory circuit also includes input lines connected to the source terminal of the memory transistor in each cell, output lines connected to an output terminal of the memristor in each cell, and programming lines coupled to a gate terminal of the memory transistor in each cell. The memristor of each cell is configured to store a conductance value representative of a synaptic weight of a synapse connected to a neuron in the artificial neural network, and the memory transistor of each cell is configured to store a threshold voltage representative of a synaptic importance value of the synapse connected to the neuron in the artificial neural network.

A nonvolatile memory device according to an embodiment includes a substrate having an upper surface, a gate line structure disposed over the substrate, a gate dielectric layer covering one sidewall surface of the gate line structure and disposed over the substrate, a channel layer disposed to cover the gate dielectric layer and disposed over the substrate, a bit line structure and a resistance change structure to contact different portions of the channel layer over the substrate, and a source line structure disposed in the resistance change structure. The gate line structure includes at least one gate electrode layer pattern and interlayer insulation layer pattern that are alternately stacked along a first direction perpendicular to the substrate, and extends in a second direction perpendicular to the first direction.

A memory device includes a memory array including a plurality of memory cells arranged in rows and columns. A closed loop bias generator is configured to output a column select signal to the memory array. A current limiter receives an output of the closed loop bias generator. The current limiter is coupled to a plurality of the columns of the memory array.

A neuromorphic computing device includes first and second memory cell arrays, and an analog-to-digital converting circuit. The first memory cell array includes a plurality of resistive memory cells, generates a plurality of read currents based on a plurality of input signals and a plurality of data, and outputs the plurality of read currents through a plurality of bitlines or source lines. The second memory cell array includes a plurality of reference resistive memory cells and an offset resistor, and outputs a reference current through a reference bitline or a reference source line. The analog-to-digital converting circuit converts the plurality of read currents into a plurality of digital signals based on the reference current. The offset resistor is connected between the reference bitline and the reference source line.

The present disclosure relates to resistive random-access memory (RRAM) devices. In some embodiments, a RRAM device may include a first electrode; a second electrode comprising an alloy containing tantalum; and a switching oxide layer positioned between the first electrode and the second electrode, wherein the switching oxide layer includes at least one transition metal oxide. The alloy containing tantalum may further contain at least one of hafnium, molybdenum, tungsten, niobium, or zirconium. In some embodiments, the alloy containing tantalum may include one or more of a binary alloy containing tantalum, a ternary alloy containing tantalum, a quaternary alloy containing tantalum, a quinary alloy containing tantalum, a senary alloy containing tantalum, and a high order alloy containing tantalum.

The present disclosure relates to resistive random-access memory (RRAM) devices. In some embodiments, an RRAM device includes: a first electrode including a metal nitride; a second electrode comprising a first conductive material; and a switching oxide layer positioned between the first electrode and the second electrode. The switching oxide layer includes at least one transition metal oxide. In some embodiments, the metal nitride in the first electrode includes titanium nitride and/or tantalum nitride. The first electrode does not include a non-reactive metal, such as platinum (Pt), palladium (Pd), etc.

Provided is a variable resistance random-access memory for suppressing degradation of performance by recovering a memory cell that fails. A variable resistance random-access memory of the disclosure includes a memory array, a row selection circuit, a column selection circuit, a controller, an error checking and correcting (ECC) circuit, an error bit flag register, and an error bit address register. The memory array includes a plurality of memory cells. The column selection circuit includes a sense amplifier and a write driver/read bias circuit. The error bit flag register stores bits for indicating presence/absence of an error bit in a write operation. The error bit address register stores an address of the error bit. The controller recovers the error bit when a predetermined event occurs.