Distribution of data in a neural network data set is used to determine an optimal compressor configuration for compressing the neural network data set and/or the underlying data type of the neural network data set. By using a generalizable optimization of examining the data prior to compressor invocation, the example non-limiting technology herein makes it possible to tune a compressor to better target the incoming data. For sparse data compression, this step may involve examining the distribution of data (e.g., in one example, zeros in the data). For other algorithms, it may involve other types of inspection. This changes the fundamental behavior of the compressor itself. By inspecting the distribution of data (e.g., zeros in the data), it also possible to very accurately predict the data width of the underlying data. This is useful because this data type is not always known a priori, and lossy compression algorithms useful for deep learning depend on knowing the true data type to achieve good compression rates.

A compression algorithm based on Huffman coding is disclosed that is adapted to be readily implemented using VLSI design. A data file may be processed to replace duplicate data with a copy commands including an offset and length, such as according to the LV algorithm. A Huffman code may then be generated for parts of the file. The Huffman code may be generated according to a novel method that generates Huffman code lengths for literals in a data file without first sorting the literal statistics. The Huffman code lengths may be constrained to be no longer than a maximum length and the Huffman code may be modified to provide an acceptable overflow probability and be in canonical order. Literals, offsets, and lengths may be separately encoded. The different values for these data sets may be assigned to a limited number of bins for purpose of generating usage statistics used for generating Huffman codes.

Described is an apparatus and method for data compression using compressive sensing in a wearable device. Described is also a machine-readable storage media having instruction stored thereon, that when executed, cause one or more processors to perform an operation comprising: receive an input signal from a sensor; convert the input signal to a digital stream; and symmetrically pad on either ends of the digital stream with a portion of the digital stream to form a padded digital stream.

A method performed in real-time includes receiving and storing time-based data over a specific time period and dividing the specific time period into a plurality of time windows. The method further includes determining that data associated with two or more proximate time windows are within a predetermined variance of one another and responsive to the determination: generating a mathematical function representative of the data associated with the two or more proximate time windows, deleting the data associated with the two or more proximate time windows, and generating a representation of the deleted data from the mathematical function. In certain embodiments, the data comprises empirical network telemetry data.

An embodiment may include circuitry that may be capable of performing compression-related operations that may include: (a) indicating, at least in part, in a data structure at least one position of at least one subset of characters that are to be encoded as a symbol, (b) comparing, at least in part, at least one pair of multi-byte data words that are of identical predetermined fixed size, (c) maintaining, at least in part, an array of pointers to potentially matching strings that are to be compared with at least one currently examined string, and/or (d) allocating, at least in part, a first buffer portion to store at least one portion of uncompressed data from an application buffer that is to be input for compression to produce a compressed data stream.

A computer-implemented method, system, and apparatus for storing binary data is disclosed. A processor receives a digital bit stream and transforms the digital bit stream to an encoded digital bit stream. The encoded digital bit stream comprises a data message encoded by an OCTS-expanded table for storage. The processor stores the encoded digital bit stream on a digital data storage device or system.

An exemplary computing environment having a DNN module can maintain one or more bandwidth throttling mechanisms. Illustratively, a first throttling mechanism can specify the number of cycles to wait between transactions on a cooperating fabric component (e.g., data bus). Illustratively, a second throttling mechanism can be a transaction count limiter that operatively sets a threshold of a number of transactions to be processed during a given transaction sequence and limits the number of transactions such as multiple transactions in flight to not exceed the set threshold. In an illustrative operation, in executing these two exemplary calculated throttling parameters, the average bandwidth usage and the peak bandwidth usage can be limited. Operatively, with this fabric bandwidth control, the processing units of the DNN are optimized to process data across each transaction cycle resulting in enhanced processing and lower power consumption.

Optimized memory usage and management is crucial to the overall performance of a neural network (NN) or deep neural network (DNN) computing environment. Using various characteristics of the input data dimension, an apportionment sequence is calculated for the input data to be processed by the NN or DNN that optimizes the efficient use of the local and external memory components. The apportionment sequence can describe how to parcel the input data (and its associated processing parameterse.g., processing weights) into one or more portions as well as how such portions of input data (and its associated processing parameters) are passed between the local memory, external memory, and processing unit components of the NN or DNN. Additionally, the apportionment sequence can include instructions to store generated output data in the local and/or external memory components so as to optimize the efficient use of the local and/or external memory components.

The performance of a neural network (NN) and/or deep neural network (DNN) can limited by the number of operations being performed as well as management of data among the various memory components of the NN/DNN. Using a directed line buffer that operatively inserts one or more shifting bits in data blocks to be processed, data read/writes to the line buffer can be optimized for processing by the NN/DNN thereby enhancing the overall performance of a NN/DNN. Operatively, an operations controller and/or iterator can generate one or more instructions having a calculated shifting bit(s) for communication to the line buffer. Illustratively, the shifting bit(s) can be calculated using various characteristics of the input data as well as the NN/DNN inclusive of the data dimensions. The line buffer can read data for processing, insert the shifting bits and write the data in the line buffer for subsequent processing by cooperating processing unit(s).

The performance of a neural network (NN) and/or deep neural network (DNN) can limited by the number of operations being performed as well as memory data management of a NN/DNN. Using vector quantization of neuron weight values, the processing of data by neurons can be optimize the number of operations as well as memory utilization to enhance the overall performance of a NN/DNN. Operatively, one or more contiguous segments of weight values can be converted into one or more vectors of arbitrary length and each of the one or more vectors can be assigned an index. The generated indexes can be stored in an exemplary vector quantization lookup table and retrieved by exemplary fast weight lookup hardware at run time on the flyas part of an exemplary data processing function of the NN as part of an inline de-quantization operation to obtain needed one or more neuron weight values.