ARITHMETIC PROCESSING DEVICE, LEARNING PROGRAM, AND LEARNING METHOD

• US 20200134434A1
• Filed: 10/24/2019
• Published: 04/30/2020
• Est. Priority Date: 10/25/2018
• Status: Abandoned Application

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Notsu, Takahiro, Kanemori, Wataru
• Application Number:
  US16/662,119
• Publication Number:
  US 20200134434A1
• Time in Patent Office: Days
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06F 7/5443 Sum of products for applications thereof, see the relevant places, e.g. G06F17/10, H03H17/00
• CPC Class Codes: G06N 3/045 Combinations of networks
• CPC Class Codes: G06N 3/048 Activation functions
• CPC Class Codes: G06N 3/063 using electronic means
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent

Abstract

An arithmetic processing device includes an arithmetic circuit; a register storing operation output data; a statistics acquisition circuit generating, from subject data being either the operation output data or normalization subject data, a bit pattern indicating a position of a leftmost set bit for positive number or a position of a leftmost zero bit for negative number of the subject data, the leftmost bit being a bit different from a sign bit; and a statistics aggregation circuit generating either positive or negative statistical information, or both positive and negative statistical information, by separately adding up a first number at respective bit positions of the leftmost set bit indicated by the bit pattern of each of a plurality of subject data having a positive sign bit and a second number of at respective bit positions of the leftmost zero bit indicated by the bit pattern of each of a plurality of subject data having a negative sign bit.

Claims

1. An arithmetic processing device comprising:

an arithmetic circuit;

a register which stores operation output data that is output by the arithmetic circuit;

a statistics acquisition circuit which generates, from subject data that is either the operation output data or normalization subject data, a bit pattern indicating a position of a leftmost set bit for positive number or a position of a leftmost zero bit for negative number of the subject data; and

a statistics aggregation circuit which generates either positive statistical information or negative statistical information, or both positive and negative statistical information, by separately adding up a first number at respective bit positions of the leftmost set bit indicated by the bit pattern of each of a plurality of subject data having a positive sign bit and a second number of at respective bit positions of leftmost zero bit indicated by the bit pattern of each of a plurality of subject data having a negative sign bit.

2. The arithmetic processing device according to claim 1, wherein the statistics aggregation circuit generates positive and negative total statistical information by adding up a third number at respective bit positions of the leftmost set bit for positive number or a position of a leftmost zero bit for negative number indicated by the bit pattern of each of a plurality of subject data having a positive sign bit and a plurality of subject data having a negative sign bit.

3. The arithmetic processing device according to claim 1, wherein the statistics aggregation circuit generates the positive statistical information or the negative statistical information by adding up the first number or the second number on the basis of a control bit indicating either a positive sign bit or a negative sign bit.

4. The arithmetic processing device according to claim 1, wherein the arithmetic circuit determines multiplication values by multiplying input data that is input respectively into a plurality of nodes in an input layer of a deep neural network by weights of edges corresponding to the nodes between the input layer and an output layer, and calculates the operation output data for each of a plurality of nodes in the output layer by cumulatively adding the multiplication values,

the statistics aggregation circuit generates the bit pattern of the operation output data calculated by the arithmetic circuit, and

the arithmetic circuit stores the operation output data in the register.

5. The arithmetic processing device according to claim 1, wherein the arithmetic circuit calculates a mean value of the operation output data on the basis of the first number, the second number, and approximate values corresponding to the position of the leftmost set bit for positive number or a position of a leftmost zero bit for negative number of the operation output data.

6. The arithmetic processing device according to claim 5, wherein the arithmetic circuit calculates a variance value of the operation output data on the basis of the approximate values of the operation output data and the mean value.

7. The arithmetic processing device according to claim 6, wherein the arithmetic circuit performs a normalization operation on the operation output data by subtracting the mean value from the operation output data and dividing the subtracted value by a square root of the variance value.

8. The arithmetic processing device according to claim 1, wherein the arithmetic circuit calculates a mean value of the normalization subject data on the basis of the first number, the second number, and approximate values corresponding to the position of the leftmost set bit for positive number or a position of a leftmost zero bit for negative number of the normalization subject data.

9. The arithmetic processing device according to claim 8, wherein the arithmetic circuit calculates a variance value of the normalization subject data on the basis of the approximate values of the normalization subject data and the mean value.

10. The arithmetic processing device according to claim 9, wherein the arithmetic circuit performs a normalization operation on the normalization subject data by subtracting the mean value from the normalization subject data and dividing the subtracted value by a square root of the variance value.

11. A non-transitory computer-readable storage medium storing therein a learning program for causing a computer to execute a learning process in a deep neural network, the learning process comprising:

reading, from a memory, statistical data of a histogram having, as a number of respective bins, a number at respective bit positions of a leftmost set bit for positive number or a position of a leftmost zero bit for negative number within subject data that is either a plurality of operation output data output by an arithmetic circuit or a plurality of normalization subject data,

calculating a mean value and a variance value of the subject data on the basis of the number of the respective bins, and approximate values each corresponding to the position of the leftmost set bit for positive number or a position of a leftmost zero bit for negative number of the subject data, and

performing a normalization operation on the subject data on the basis of the mean value and the variance value.

12. A learning method for causing a processor to execute a learning process in a deep neural network, the learning process comprising:

reading, from a memory, statistical data of a histogram having, as a number of respective bins, a number at respective bit positions of a leftmost set bit for positive number or a position of a leftmost zero bit for negative number within subject data that is either a plurality of operation output data output by an arithmetic circuit or normalization subject data,

calculating a mean value and a variance value of the subject data on the basis of the number of the respective bins and approximate values each corresponding to the position of the leftmost set bit for positive number or a position of a leftmost zero bit for negative number of the subject data, and

performing a normalization operation on the subject data on the basis of the mean value and the variance value.