A method for differentiator-based compression of digital data includes (a) multiplying a tap-weight vector by an original data vector to generate a predicted signal, the original data vector comprising N sequential samples of an original signal, N being an integer greater than or equal to one, (b) using a subtraction module, subtracting the predicted signal from a sample of the original signal to obtain an error signal, (c) using a quantization module, quantizing the error signal to obtain a quantized error signal, and (d) updating the tap-weight vector according to changing statistical properties of the original signal.

An optical network includes a transmitting portion configured to (i) encode an input digitized sequence of data samples into a quantized sequence of data samples having a first number of digits per sample, (ii) map the quantized sequence of data samples into a compressed sequence of data samples having a second number of digits per sample, the second number being lower than the first number, and (iii) modulate the compressed sequence of data samples and transmit the modulated sequence over a digital optical link. The optical network further includes a receiving portion configured to (i) receive and demodulate the modulated sequence from the digital optical link, (ii) map the demodulated sequence from the second number of digits per sample into a decompressed sequence having the first number of digits per sample, and (iii) decode the decompressed sequence.

A delta sigma modulator includes a summation circuit, at least one integrator, a multi-bit quantizer and a negative feedback circuit. The summation circuit is configured to produce a difference signal between a unipolar or bipolar analog input signal and an analog feedback signal. The integrator is operatively coupled to the summation circuit to integrate the difference signal. The multi-bit quantizer is operatively coupled to the integrator to digitize the integrated signal to generate an N-bit digital output signal, N being an integer greater than 1. The negative feedback circuit operatively couples the multi-bit quantizer to the summation circuit. The negative feedback circuit includes a digital-to-analog converter arrangement for receiving the N-bit digital output signal and providing the analog feedback signal such that digital values of the N-bit digital output signal and values of the analog feedback encoded by the digital values have a non-linear relationship to one another.

An optical network includes a transmitting portion configured to (i) encode an input digitized sequence of data samples into a quantized sequence of data samples having a first number of digits per sample, (ii) map the quantized sequence of data samples into a compressed sequence of data samples having a second number of digits per sample, the second number being lower than the first number, and (iii) modulate the compressed sequence of data samples and transmit the modulated sequence over a digital optical link. The optical network further includes a receiving portion configured to (i) receive and demodulate the modulated sequence from the digital optical link, (ii) map the demodulated sequence from the second number of digits per sample into a decompressed sequence having the first number of digits per sample, and (iii) decode the decompressed sequence.

An electronic device may include an electronic display having multiple display pixels to display an image based on analog voltage signals. The electronic device may also include optical calibration circuitry to generate digital-to-analog converter (DAC) data based on image data associated with the image and dither circuitry to reduce a bit-depth of the DAC data, generating dithered DAC data. Additionally, the electronic device may include a gamma generator having one or more DACs to generate the analog voltage signals based on the dithered DAC data, which may instruct the gamma generator to generate the analog voltage signals indicative of the image data.

The present disclosure provides digital to analog conversion circuitry comprising: a set of input nodes for receiving a digital input code; an output node for outputting an analog output signal representative of the input code; and a plurality of selectable conversion elements, wherein a parameter of each of the plurality of selectable conversion elements is configured such that a transfer function between the input code and the output analog signal is non-monotonic.

The present disclosure relates to a compensation circuit for compensating for an offset voltage that is present in an output signal output by a force sensor. The compensation circuit comprises: voltage divider circuitry, the voltage divider circuitry configured to receive a bias voltage that is also supplied to the force sensor and to output a control voltage derived from the bias voltage, wherein a component mismatch ratio of the voltage divider circuitry is adjustable to correspond to a component mismatch ratio of the force sensor; current generator circuitry configured to receive the control voltage and to generate a compensating current based on the received control voltage; and amplifier circuitry configured to receive the differential signal output by the force sensor and the compensating current and to output a compensated differential output signal in which the offset voltage is at least partially cancelled.

The present disclosure provides digital to analog conversion circuitry comprising: a set of input nodes for receiving a digital input code; an output node for outputting an analog output signal representative of the input code; and a plurality of selectable conversion elements, wherein a parameter of each of the plurality of selectable conversion elements is configured such that a transfer function between the input code and the output analog signal is non-monotonic.

A method for differentiator-based compression of digital data includes (a) multiplying a tap-weight vector by an original data vector to generate a predicted signal, the original data vector comprising N sequential samples of an original signal, N being an integer greater than or equal to one, (b) using a subtraction module, subtracting the predicted signal from a sample of the original signal to obtain an error signal, (c) using a quantization module, quantizing the error signal to obtain a quantized error signal, and (d) updating the tap-weight vector according to changing statistical properties of the original signal.

Methods of performing mixed-signal current-mode multiply-accumulate (MAC) operations for binarized neural networks in an integrated circuit are described in this disclosure. While digital machine learning circuits are fast, scalable, and programmable, they typically require bleeding-edge deep sub-micron manufacturing, consume high currents, and they reside in the cloud, which can exhibit long latency, and not meet private and safety requirements of some applications. Digital machine learning circuits also tend to be pricy given that machine learning digital chips typically require expensive tooling and wafer fabrication associated with advanced bleeding-edge deep sub-micron semiconductor manufacturing. This disclosure utilizes mixed-signal current mode signal processing for machine learning binarized neural networks (BNN), including Compute-In-Memory (CIM), which can enable on-or-near-device machine learning and or on sensor machine learning chips to operate more privately, more securely, with low power and low latency, asynchronously, and be manufacturable on non-advanced standard sub-micron fabrication (with node portability), that are more mature and rugged with lower costs. An example of enabling features of this disclosure is as follows: to save power in an always-on setting, reduce chip costs, process signals asynchronously, and reduce dynamic power consumption. Current mode signal processing is utilized in combination with CIM (to further reduce dynamic power consumption associated with read/write cycles in and out of memory) for bitwise counting of plurality of logic state 1 of plurality of XOR outputs for MAC arithmetic in BNNs.