A processing device for executing a machine learning neural network operation includes memory and a processor. The processor is configured to receive input data at a layer of the machine learning neural network operation, receive a plurality of sorted filters to be applied to the input data, apply the plurality of sorted filters to the input data to produce a plurality of different feature maps, compress the plurality of different feature maps according to a similarity of the feature maps relative to each other and store the plurality of different feature maps in the memory.

A device includes a sensor configured to output an analog sensor signal, an analog-to-digital converter circuit configured to convert the analog sensor signal into a sigma-delta-modulated digital signal having a bit width of n bits, and a pulse width modulator configured to generate a pulse-width-modulated signal based on the sigma-delta-modulated digital signal.

The invention provides a system improving signal handling, e.g., transmission and/or processing. In an embodiment, the system may include a filter circuit, a magnitude bit truncation circuit and a utility circuit. The filter circuit may be coupled to a target signal which contains one or more desired signals at one or more interested bands, for attenuating each said interested band to form a filtered signal. The magnitude bit truncation circuit may be coupled to the filter circuit, for truncating one or more bits of each sample of the filtered signal to form a truncated signal. The utility circuit may be coupled to the magnitude bit truncation circuit, for handling the truncated signal to implement handling of the target signal, so as to reduce resource requirement and enhance error tolerance comparing with directly handling the target signal.

An apparatus for mitigating nonlinearity-induced spurs and noise in a fractional-N frequency synthesizer

A digital delta-sigma modulator (DDSM) is disclosed with an input signal x[n], an output signal y[n], a quantization error signal e[n] and a dither signal d[n], having an equation described in the z-domain by

Y(z)=STF(z)X(z)+DTF(z)D(z)−NTF(z)E(z)

wherein Y(z), X(z), D(z) andE(z) are z-transforms of the output signal, the input signal, the dither signal, and the quantization error signal, and wherein STF (z), DTF(z) and NTF(z) correspond to a transfer function of the input signal, a transfer function of the dither signal, and a transfer function of the quantization error signal, and wherein the transfer function of the quantization error signal is of the form:

[00001]$\mathrm{NTF}\left(z\right)={\mathrm{Az}}^{-Q}\left(1-z^{-1}\right)\left(1+\underset{i=1}{\overset{K}{.Math.}}{c}_i{z}^{-i}\right)$

where A , Q and K are constants, coefficients c.sub.i are real valued and c.sub.K≠0 and wherein at least one of the zeroes z.sub.j of

[00002]$\left(1+\underset{i=1}{\overset{K}{.Math.}}{c}_i{z}^{-i}\right)$

satisfies z.sub.j≠+1 for j=1, 2, . . . , K

A transducer system has 1) a MEMS transducer configured to produce an analog output signal in response to an incident acoustic signal, and 2) a modulator apparatus. The modulator apparatus includes an analog modulator configured to receive the analog output signal and produce a first digital signal as a function of the analog output signal. In addition, the modulator apparatus also has a digital converter configured to receive the first digital signal and produce a second digital signal as a function of the first digital signal. The analog modulator has an analog order while, in a corresponding manner, the digital converter has a digital order. Preferably, the digital order is higher than the analog order.

Systems and methods for a power-efficient 3-level digital-to-analog converter. A converter cell using a current starving technique keeps a portion of the converter cell turned on in a low power mode, as opposed to completely turning off current in selected modes. A conversion system keeps a first set of converters active while allowing a second set of converters to be powered down. Systems and methods presented save power and allow for efficient reactivation of converters.

Methods are provided for efficiently encoding and decoding multivariate correlated data sequences for transmission over multiple channels of a network. The methods include transforming data vectors from correlated sources into vectors that comprise substantially independent and correlated components, and generating a common information vector based on the correlated components, and two private information vectors. The methods also include computing the amount of information, such as Wyner's lossy common information, in the common information vector, computing rates that lie on the Gray-Wyner rate region, and choosing compression rates based on the amount of common information. The methods may be applicable, in general, to a wide range of communications and/or storage systems and, particularly, to sensor networks and multi-user virtual environments for gaming and other applications.

A method for managing data includes obtaining a compression algorithm selection request for a data object, wherein the data object is generated by a production host, identifying, in response to the compression algorithm selection request, a set of production host performance objectives of the production host, performing a compression algorithm selection analysis using the set of production host performance objectives and a compression selection model to obtain a compression algorithm selection for a compression algorithm, specifying the compression algorithm to the production host using a data agent, wherein the data agent is operatively connected to the production host, initiating a compression on the data object using the data agent by applying the compression algorithm to obtain a compressed data object, and initiating a storage of the compressed data object.

The method involves sequentially encoding the plurality of data packets, which involves, for a data packet, obtaining a reference data packet, obtaining a mask packet indicative of which of the bits in the data packet are predictable and which of the bits in the data packet are not predictable obtaining change history data indicative of previously changed bits of the mask packet, determining an updated mask packet based on the mask packet, the data packet, and the reference data packet, determining updated change history data based on the change history data, the mask packet, and the updated mask packet, determining, as unpredictable bits, all those bits of the data packet that are indicated as not predictable by the updated mask packet, and generating an encoded data packet including a representation of the updated change history data and a representation of values of the unpredictable bits of the data packet.

Systems and methods for improving noise efficiency in a Delta Sigma modulator. A bypass scheme for a noise splitter is disclosed that reduces toggling activity for small signals. In particular, a sample-by-sample bypass noise splitter is disclosed that includes a noise splitting module and a bypass line. The bypass line bypasses the noise splitting module when signals are below a selected threshold, increasing efficiency of the system.