A method for adapting a trained neural network is provided. Input data is input to the trained neural network and a plurality of filters are applied to generate a plurality of channels of activation data. Differences between corresponding activation values in the plurality of channels of activation data are calculated and an order of the plurality of channels is determined based on the calculated differences. The neural network is adapted so that it will output channels of activation data in the determined order. The ordering of the channels of activation data is subsequently used to compress activation data values by taking advantage of a correlation between activation data values in adjacent channels.

A system and method for compacting data that uses mismatch probability estimation to improve entropy encoding methods to account for, and efficiently handle, previously-unseen data in data to be compacted. Training data sets are analyzed to determine the frequency of occurrence of each sourceblock in the training data sets. A mismatch probability estimate is calculated comprising an estimated frequency at which any given data sourceblock received during encoding will not have a codeword in the codebook. Entropy encoding is used to generate codebooks comprising codewords for data sourceblocks based on the frequency of occurrence of each sourceblock. A “mismatch codeword” is inserted into the codebook based on the mismatch probability estimate to represent those cases when a block of data to be encoded does not have a codeword in the codebook. During encoding, if a mismatch occurs, a secondary encoding process is used to encode the mismatched sourceblock.

A computer-implemented method reduces an amount of machine health data to be stored in a data storage device while preserving details of extrema values occurring within incremental measurement time intervals in an extended time period during which the data were collected. The method includes: sensing an operational characteristic of a machine and generating an operational characteristic signal; generating machine health parameter data that include amplitude values and associated time values; for each incremental measurement time interval, calculating an average value of the amplitude values, identifying a maximum value of the amplitude values, and identifying a minimum value of the amplitude values; and storing a compressed machine health parameter data set in the data storage device. The compressed machine health parameter data set includes the calculated average values and the identified maximum and minimum values for the incremental measurement time intervals.

The invention describes an encoding scheme for arithmetically encoding a sequence of information values into an arithmetic coded bitstream using providing the bitstream with entry point information allowing for resuming arithmetic decoding the bitstream from a predetermined entry point onward. A respective decoding scheme is also provided. These encoding and decoding schemes provide more efficient encoding concept in view of the decoding speed.

A method and device for compressing FST data are provided. The method includes: acquiring to-be-compressed FST data, where the FST data includes state transition data and state data; decomposing the state transition data based on first data categories to acquire first decomposition data; decomposing the state data based on second data categories to acquire second decomposition data; sequentially arranging, for each of the first data categories, the first decomposition data of the first data category, to acquire first arrangement data of the first data category; alternately arranging the first arrangement data and the second decomposition data according to a sequential order used in the first arrangement data, to acquire second arrangement data; performing classification statistics on the first arrangement data and the second arrangement data to acquire index data; and combining the first arrangement data, the second arrangement data, and the index data, to obtain the compressed FST data.

Technologies for compressing a blockchain. In some examples, the technologies include removing selected blocks within a blockchain, and replacing the selected blocks with a summary block and a padding block. Each block of the selected blocks includes data in a certain state (such as data in an obsolete state). The technologies can include generating the summary block and padding blocks according to the data in the selected blocks and an original root hash included in the selected blocks and other blocks of the blockchain. The generating of the summary and padding blocks can include generating a new root hash in the summary and padding blocks that only replaces the original root hash in the summary and padding blocks. The generation of the new root hash can be based on a part of a header of a non-selected block of the blockchain directly linked to an end block of selected blocks.

A device configured to compress a tensor including a plurality of cells includes: a quadtree generator configured to generate a quadtree searching for a non-zero cell included in the tensor and extract at least one parameter value from the quadtree; a mode selector configured to determine a compression mode based on the at least one parameter; and a bitstream generator configured to generate a bitstream by compressing the tensor based on the compression mode.

A system can identify a first group of time-series telemetry data that represents performance metrics of a computing device, wherein the first group of time-series telemetry data identifies respective first values and corresponding respective first timestamps. The system can create a second group of time-series telemetry data that identifies second timestamps. The system can populate the second group of time-series telemetry data with the respective first values at respective first locations of the second group of time-series telemetry data that correspond to the respective first timestamps of the respective first values. The system can create a tensor that identifies third timestamps. The system can populate the tensor with the respective first values at respective second locations of the tensor that correspond to the respective first timestamps of the respective first values, wherein populating the tensor comprises combining two values of the respective first values.

The present disclosure provides an integrated circuit chip device and a related product. The integrated circuit chip device includes: a primary processing circuit and a plurality of basic processing circuits. The primary processing circuit or at least one of the plurality of basic processing circuits includes the compression mapping circuits configured to perform compression on each data of a neural network operation. The technical solution provided by the present disclosure has the advantages of a small amount of computations and low power consumption.

Provided is a method for encoding scatterplot data using strings. The method may comprise receiving a plurality of data points in a data set. Each data point has at least a first data value corresponding to a first dimension and a second data value corresponding to a second dimension. The method further comprises determining a first resolution for the first dimension and a second resolution for the second dimension. The method further comprises determining an encoding scheme for encoding the plurality of data points. The encoding scheme includes a plurality of valid encoding characters. The method further comprises encoding each of the plurality of data points based on the first resolution, the second resolution, and the encoding scheme.