A computer-implemented method for improving compression of predictive models includes generating an unlabeled simulated data set by expanding an initial data set, and generating a labeled data set by predicting the unlabeled, simulated data set using a complex model to output a plurality of labels. The method also includes training a relatively simple neural network using the labeled data set.

A memory system includes a nonvolatile memory, an interface circuit, and a controller configured to upon receipt of a plurality of write commands for storing write data in the nonvolatile memory via the interface circuit, acquire compression-ratio information about the write data associated with each write command, determine a compression ratio of each write data based on the acquired compression-ratio information, and determine an execution order of the write commands based on the determined compression ratio.

Systems and methods are disclosed for improving data channel design by applying transform domain analytics to more reliably extract user data from a signal. In certain embodiments, an apparatus may comprise a channel circuit configured to receive an analog signal at an input of the channel circuit, and sample the analog signal to obtain a set of signal samples. The channel circuit may further apply a filter configured to perform transform domain analysis to the set of signal samples to generate a first subset of samples, the first subset including fewer transitions and having a higher signal to noise ratio (SNR) than the set of signal samples. The channel circuit may detect first bit transform domain representation values from the first subset, and determine channel bit values encoded in the analog signal based on the set of signal samples and using the first bit transform domain representation values detected from the first subset as side information.

Systems and methods for fault-tolerant threshold circuits used in converting an analog input to a single-bit digital output employ N-modular redundancy of either inverting or non-inverting threshold circuits whose inputs are connected to a single input, and apply majority voting of their outputs to provide correction of transient or permanent faults in up to floor[(N−1)/2] of the individual threshold circuits. Using summation to perform analog majority voting averages the N threshold circuit outputs and provides resilience to single-event transients, but may exhibit an output characteristic having intermediate voltage levels. A digital majority voter having N inputs connected to the outputs of N threshold circuits restores well-defined logic levels and clean hysteresis for Schmitt trigger threshold circuits. A single point of failure at the digital majority voter may be eliminated using an analog majority voter to sum the outputs of three or more redundant digital majority voters.

In general, embodiments of the present invention provide systems, methods and computer readable media for data record compression using graph-based techniques. An example method includes determining a plurality of index components; generating a sorted data record list of a plurality of compound data records; generating an ordered unique index component value list associated with a plurality of unique index component values; assigning a plurality of encodings to the plurality of unique index component values; and generating the compressed data record list based on the ordered unique index component value list and the plurality of encodings.

Aspects of the disclosure provide methods and apparatuses for neural network model compression/decompression. In some examples, an apparatus for neural network model decompression includes receiving circuitry and processing circuitry. The processing circuitry decodes, from a bitstream corresponding to a representation of a neural network, at least a syntax element to be applied to multiple blocks in the neural network. Then, the processing circuitry reconstructs, from the bitstream, weight coefficients in the blocks based on the syntax element.

A decoding device is used to securely send corresponding data gathered from multiple underground sources to multiple users. The device comprises a signal receiving port connected to multiple bandwidth filters and further connected to internet access points that are assigned to end users for secure data access. The invention facilitates allowing the signal and data being transmitted through the formation of the earth to reach end users located nearby and significant distances away from the source of the transmission. A system and method utilizing the decoding device is provided.

The present invention discloses a DAC method having signal calibration mechanism. A first conversion circuit generates a first analog signal according to an input digital signal. A second conversion circuit generates a second analog signal according to the input digital signal and a pseudo-noise digital signal. An echo transmission circuit processes a signal on an echo path to generate an echo signal. A first and a second calibration circuits generate a first and a second calibration signals. A calibration parameter calculation circuit performs calculation according to a difference between the echo signal and a sum of the first and the second calibration signals and related path information to generate a first and a second offsets. The first and the second calibration circuits converge first and second response coefficients and update a first and a second codeword offset tables according to the first and the second offsets.

A system includes a non-transitory computer readable medium configured to store instructions thereon; and a processor connected to the non-transitory computer readable medium. The processor is configured to execute the instructions for generating a mask based on received data from a sensor, wherein the mask includes a plurality of importance values, and each region of the received data is designated a corresponding importance value of the plurality of importance values. The processor is configured to execute the instructions for encoding the received data based on the mask; and transmitting the encoded data to a decoder for defining reconstructed data. The processor is configured to execute the instructions for computing a loss based on the reconstructed data, the received data and the mask. The processor is configured to execute the instructions for providing training to an encoder for encoding the received data based on the computed loss.

Methods, systems, and apparatus, including computer-readable storage media for hardware compression and decompression. A system can include a decompressor device coupled to a memory device and a processor. The decompressor device can be configured to receive, from the memory device, compressed data that has been compressed using an entropy encoding, process the compressed data using the entropy encoding to generate uncompressed data, and send the uncompressed data to the processor. The system can also include a compressor device configured to generate, from uncompressed data, a probability distribution of codewords, generate a code table from the probability distribution, and compress incoming data using the generated code table.