A method of generating a fixed-point quantized neural network includes analyzing a statistical distribution for each channel of floating-point parameter values of feature maps and a kernel for each channel from data of a pre-trained floating-point neural network, determining a fixed-point expression of each of the parameters for each channel statistically covering a distribution range of the floating-point parameter values based on the statistical distribution for each channel, determining fractional lengths of a bias and a weight for each channel among the parameters of the fixed-point expression for each channel based on a result of performing a convolution operation, and generating a fixed-point quantized neural network in which the bias and the weight for each channel have the determined fractional lengths.

A method of generating a fixed-point quantized neural network includes analyzing a statistical distribution for each channel of floating-point parameter values of feature maps and a kernel for each channel from data of a pre-trained floating-point neural network, determining a fixed-point expression of each of the parameters for each channel statistically covering a distribution range of the floating-point parameter values based on the statistical distribution for each channel, determining fractional lengths of a bias and a weight for each channel among the parameters of the fixed-point expression for each channel based on a result of performing a convolution operation, and generating a fixed-point quantized neural network in which the bias and the weight for each channel have the determined fractional lengths.

The present disclosure relates to a converter for data type conversion, a method for data type conversion, an integrated circuit chip, and a calculation apparatus, where the calculation apparatus may be included in a combined processing apparatus, where the combined processing apparatus may further include a general interconnection interface and other processing apparatus. The calculation apparatus interacts with other processing apparatus to jointly complete calculation operations specified by users. The combined processing apparatus may further include a storage apparatus. The storage apparatus is respectively connected to the calculation apparatus and other processing apparatus, and the storage apparatus is used for storing data of the calculation apparatus and other processing apparatus. A solution of the present disclosure may be widely applied to various data type conversion applications.

In mobile communications networks, requirements on signal distortion may be fulfilled at a lower bit rate, or alternatively quantization noise be reduced for a given bit rate, by including fractional exponent bits in a block floating point format. One or more fractional exponent bits may apply to all samples in the block. Alternatively, fractional bits may apply to sub-blocks within the block. The optimal number of fractional bits depends on the number of samples in the block.

In mobile communications networks, requirements on signal distortion may be fulfilled at a lower bit rate, or alternatively quantization noise be reduced for a given bit rate, by including fractional exponent bits in a block floating point format. One or more fractional exponent bits may apply to all samples in the block. Alternatively, fractional bits may apply to sub-blocks within the block. The optimal number of fractional bits depends on the number of samples in the block.

The present disclosure provides a computation device. The computation device is configured to perform a machine learning computation, and includes an operation unit, a controller unit, and a conversion unit. The storage unit is configured to obtain input data and a computation instruction. The controller unit is configured to extract and parse the computation instruction from the storage unit to obtain one or more operation instructions, and to send the one or more operation instructions and the input data to the operation unit. The operation unit is configured to perform operations on the input data according to one or more operation instructions to obtain a computation result of the computation instruction. In the examples of the present disclosure, the input data involved in machine learning computations is represented by fixed-point data, thereby improving the processing speed and efficiency of training operations.

The present disclosure provides a computation device. The computation device is configured to perform a machine learning computation, and includes an operation unit, a controller unit, and a conversion unit. The storage unit is configured to obtain input data and a computation instruction. The controller unit is configured to extract and parse the computation instruction from the storage unit to obtain one or more operation instructions, and to send the one or more operation instructions and the input data to the operation unit. The operation unit is configured to perform operations on the input data according to one or more operation instructions to obtain a computation result of the computation instruction. In the examples of the present disclosure, the input data involved in machine learning computations is represented by fixed-point data, thereby improving the processing speed and efficiency of training operations.

A MAC operator includes a plurality of multipliers configured to perform a multiplication operation on a floating-point format first data and a floating-point format second data to output a floating-point format multiplication result data, a plurality of floating-point-to-fixed-point converters configured to receive the floating-point format multiplication result data from each of the plurality of multipliers and convert into a fixed-point format multiplication result data to be output, and an adder tree configured to perform an addition operation on the fixed-point format multiplication result data that is output from the plurality of floating-point-to-fixed-point converters. If a first mantissa of the first data and a second mantissa of the second data are composed of ‘M’-bit (‘M’ being a natural number), each of the plurality of multipliers is configured to perform the multiplication operation so that the fixed-point format multiplication result data includes a mantissa of 2*(M+1) bits.

A MAC operator includes a plurality of multipliers configured to perform a multiplication operation on a floating-point format first data and a floating-point format second data to output a floating-point format multiplication result data, a plurality of floating-point-to-fixed-point converters configured to receive the floating-point format multiplication result data from each of the plurality of multipliers and convert into a fixed-point format multiplication result data to be output, and an adder tree configured to perform an addition operation on the fixed-point format multiplication result data that is output from the plurality of floating-point-to-fixed-point converters. If a first mantissa of the first data and a second mantissa of the second data are composed of ‘M’-bit (‘M’ being a natural number), each of the plurality of multipliers is configured to perform the multiplication operation so that the fixed-point format multiplication result data includes a mantissa of 2*(M+1) bits.

A method for processing data sets having at least one time series. These are measurement values of sensors that are sensed at certain times. The data of the measured values and the respective times at which the data were sensed are stored as data elements of the time series. The method compresses the data set by rounding the sensed data with subsequent decimation of the data elements of the data set which are contained in the time series.