Method and apparatus for enhancing processing speed for performing a least mean square operation by parallel processing
The present invention is related to a method and apparatus for enhancing processing speed for performing a least mean square operation by parallel processing. The input data are subdivided into a plurality of groups and processed by a plurality of adaptive filters in parallel. A Jaber product device is utilized for rearranging the processing results. A subtractor subtracts the output from the Jaber product device from a desired result to generate an error signal. A feedback network adjusts the adaptive filters in accordance with the error signal.
1. An apparatus for performing linear regression of input data, the apparatus comprising:
a plurality of multipliers for multiplying each of a plurality of subdivided input data picked up at a stride r with a corresponding weight, respectively, to generate partial linear regression outputs;
a Jaber product device for rearranging the partial linear regression outputs and generating a linear regression output; and
a subtractor for subtracting the linear regression output from a desired result to generate an error of a linear regression model.
2. The apparatus of claim 1 further comprising a second multiplier for biasing the linear regression output.
3. An apparatus for performing regression of multi-variable input data, the apparatus comprising:
a plurality of multipliers for multiplying each of a plurality of subdivided multi-variable input data with a corresponding weight, respectively, to generate partial multi-variable regression outputs;
a plurality of Jaber product devices for rearranging the partial multi-variable regression outputs to generate multi-variable regression outputs; and
a subtractor for subtracting the multi-variable regression outputs from a desired result to generate an error of a multi-variable regression model.
4. The apparatus of claim 3 further comprising a plurality of second multipliers for biasing the multi-variable partial regression outputs.
5. An apparatus for processing input data in parallel, the apparatus comprising:
a plurality of adaptive filters for processing input data in parallel, each adaptive filter accepting each of a plurality of subdivided input data and generating a partial filtered output;
a plurality of subtractors, each subtractor for subtracting the partial filtered output from a partial desired output for generating a partial error signal;
a plurality of feedback networks for adjusting the adaptive filters in accordance with a corresponding partial error signal; and
a Jaber product device for rearranging partial error signals from the adaptive filters to generate an error output.
6. A method for performing linear regression of input data, the method comprising:
receiving input data;
dividing the input data into a plurality of subgroups;
multiplying each subdivided input data with a corresponding weight, respectively, in parallel, to generate partial linear regression outputs;
rearranging the partial linear regression outputs to generate a linear regression output; and
subtracting the linear regression output from a desired result to generate an error of a linear regression model.
7. A method for processing input data in parallel, the method comprising:
receiving input data;
dividing the input data into a plurality of subgroups;
filtering each of the subdivided input data in parallel with a plurality of adaptive filters to generate partial filtered outputs;
subtracting each of the partial filtered outputs from a desired result to generate partial errors;
adjusting the adaptive filters in accordance with the partial errors; and
rearranging the partial errors to generate an error output.
8. An apparatus for performing linear regression of input data, the apparatus comprising:
a plurality of multipliers for multiplying each of a plurality of subdivided input data picked up at a stride r with corresponding weights, respectively, to generate partial linear regression outputs;
a plurality of subtractor for subtracting each of the partial linear regression outputs from a partial desired result to generate partial errors; and
a Jaber product device for rearranging the partial errors and generating an error of a linear regression model.
9. The apparatus of claim 8 further comprising a plurality of second multipliers for biasing the partial linear regression outputs.
10. An apparatus for performing regression of multi-variable input data, the apparatus comprising:
a plurality of multipliers for multiplying each of a plurality of subdivided multi-variable input data with a corresponding weight, respectively, to generate partial multi-variable regression outputs;
a plurality of subtractors for subtracting each of the partial multi-variable regression outputs from a partial desired result to generate multi-variable partial errors; and
a Jaber product device for rearranging the multi-variable partial errors to generate an error of a multi-variable regression model.
11. The apparatus of claim 10 further comprising a plurality of second multipliers for biasing the multi-variable partial regression outputs.
12. A method for performing linear regression of input data, the method comprising:
receiving input data;
dividing the input data into a plurality of subgroups;
multiplying each subdivided input data with a corresponding weight, respectively, in parallel, to generate linear partial regression outputs;
subtracting each of the linear partial regression outputs from a partial desired result to generate partial errors; and
rearranging the partial errors to generate an error of a linear regression model.