Neural processor which can calculate a norm or a distance

• US 5,537,513 A
• Filed: 06/05/1995
• Issued: 07/16/1996
• Est. Priority Date: 04/29/1992
• Status: Expired due to Fees

• Current Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Original Assignee: US Philips Corporation (Koninklijke Philips N.V.)
• Inventors: DeVille, Yannick
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Hafiz, Tariq
• Application Number:
  US08/461,474
• Time in Patent Office: 407 Days
• Field of Search: 395/20-25, 395/26-27, 382/155-159
• US Class Current: 706/41
• CPC Class Codes: G06F 18/22 Matching criteria, e.g. proximity measures
• CPC Class Codes: G06F 2207/4802 Special implementations
• CPC Class Codes: G06F 2207/5525 Pythagorean sum, i.e. the square root of a sum of squares
• CPC Class Codes: G06F 7/552 Powers or roots , e.g. Pythagorean sums
• CPC Class Codes: G06N 3/04 Architecture, e.g. interconnection topology

Abstract

A neural processor, comprising neural calculation apparatus (30, NQ, RQ) which extracts a root Q of a quantity X, said root constituting either a norm of a data or a distance between data. The calculation apparatus calculates (30) by iteration a series of contributions ΔQ_i which are used (NQ, RQ) to update a partial root QP which becomes the root Q at the end of calculation. The calculation can be performed on an arbitrary arithmetic base which determines the number of neurons utilized and also the accuracy of calculation. It is also possible to execute the calculation of a partial remainder (NR, RR). Several programming modes are presented.

Claims

1. A neural processor, comprising calculation apparatus (11-15) for extracting a root Q of a data X, which root constitutes either a norm of data or a distance between data, wherein the calculation apparatus comprises

a preliminary neuron device (30) having

an output;

at least one plurality of (B-1) neurons arranged in parallel for computing terms q_i ×

Bⁱ expressing a serial development of the root Q in an arithmetic base B, where B, q, and i are integers;

a two-input first neuron (NQ), having a first input coupled to the output of the preliminary neuron device;

a three-input second neuron, having a first input coupled to the output of the preliminary neuron device;

first delay means (RQ) for delaying a signal on an output of the first neuron, the first delay means having an output for providing, ultimately, the root Q, the output of the first delay means being recursively coupled to a second input of the first neuron, a first input of the preliminary neuron device, and to a second input of the second neuron; and

second delay means (RR) for delaying a signal on an output of the second neuron and having

a first input coupled with the output of the second neuron;

an output for providing, ultimately, a remainder R, the output of the second delay means being recursively coupled to a third input of the second neuron and to a second input of the preliminary neuron device; and

a second input for initializing the second delay means with the data X.

2. A neural processor, comprising calculation apparatus (11-15) for extracting a root Q of a data X, which root constitutes wither a norm of data or a distance between data, wherein the calculation apparatus comprises

a preliminary neuron device (30) comprising:

a first input;

a second input for receiving the data X; and

at least one plurality of (B-1) neurons arranged in parallel for computing terms q_i ×

Bⁱ expressing a serial development of the root Q in an arithmetic base B, where B, q, and i are integers;

a two-input first neuron (NQ) having a first input coupled with an output of the preliminary neuron device;

a three-input second neuron (NR), having a first input coupled with the output of the preliminary neuron device;

first delay means (RQ) for delaying a signal on an output of the first neuron, the first delay means having an output for providing, ultimately, the root Q, the output of the first delay means being recursively connected to a second input of the first neuron, a first input of the preliminary neuron device, and a second input of the second neuron;

second delay means (RR) for delaying a signal on an output of the second neuron and having

a first input coupled to an output of the second neuron;

an output for providing, ultimately, a remainder R, the output of the second delay means being recursively connected to a third input of the second neuron; and

a second input for initializing the second delay means with the data X.

3. A neural processor, comprising calculation apparatus (11-15) for extracting a root Q of a data X, which root constitutes either a norm of data or a distance between data, wherein the calculation apparatus comprises

a preliminary neuron device (30) having

a first input;

a second input for receiving the data X; and

at least one plurality of (B-1) neurons arranged in parallel for computing terms q_i ×

Bⁱ expressing a serial development of the root Q in an arithmetic base B, where B, q, and i are integers;

a two-input first neuron (NQ) having a first input coupled to an output of the preliminary neuron device; and

first delay means for delaying a signal at an output of the first neuron, the first delay means having an output for providing, ultimately, the root Q, the output of the first delay means being connected recursively to a second input of the first neuron and the first input of the preliminary neuron device.