Feature dataset classification

Patent metadata

Field	Value
Publication number	US-12067086-B2
Application number	US-202017593716-A
Country	US
Kind code	B2
Filing date	Feb 27, 2020
Priority date	Mar 29, 2019
Publication date	Aug 20, 2024
Grant date	Aug 20, 2024

Abstract

Apparatuses and methods of operating such apparatuses are disclosed. An apparatus comprises feature dataset input circuitry to receive a feature dataset comprising multiple feature data values indicative of a set of features, wherein each feature data value is represented by a set of bits. Class retrieval circuitry is responsive to reception of the feature dataset from the feature dataset input circuitry to retrieve from class indications storage a class indication for each feature data value received in the feature dataset, wherein class indications are predetermined and stored in the class indications storage for each permutation of the set of bits for each feature. Classification output circuitry is responsive to reception of class indications from the class retrieval circuitry to determine a classification in dependence on the class indications. A predicated class may thus be accurately generated from a simple apparatus.

First claim

The invention claimed is:

1. An apparatus embodied as a plastic fabricated device, the apparatus comprising:
feature dataset input circuitry, included as part of the plastic fabricated device, to receive a feature dataset comprising multiple feature data values indicative of a set of features, wherein each feature data value is represented by a set of bits;
class retrieval circuitry, included as part of the plastic fabricated device and responsive to reception of the feature dataset from the feature dataset input circuitry, to retrieve from class indications storage a class indication for each feature data value received in the feature dataset, wherein the class indications are predetermined and stored in the class indications storage for each possible value of the set of bits for each feature, and wherein the class indications storage has a look-up table format and the class retrieval circuitry is arranged to perform a look-up procedure with respect to the look-up table format for each feature data value received in the feature dataset; and
classification output circuitry, included as part of the plastic fabricated device and responsive to a reception of the class indications from the class retrieval circuitry, to determine a classification by a vote amongst the class indications.

2. The apparatus as claimed in claim 1 , wherein the class indications stored in the class indications storage for each feature are each predetermined as a best class indication which maximises a Bayes Classifier for the feature in a training phase using feature training datasets.

3. The apparatus as claimed in claim 2 , wherein the Bayes Classifier is a Gaussian naïve Bayes Classifier.

4. The apparatus as claimed in claim 2 , wherein the Bayes Classifier is based on a single distribution type used for each feature of the set of features.

5. The apparatus as claimed in claim 2 , wherein the Bayes Classifier is based on heterogeneous distribution types used for the set of features.

6. The apparatus as claimed in claim 1 , wherein the class retrieval circuitry is arranged to retrieve in parallel the class indications for each feature data value received in the feature dataset.

7. The apparatus as claimed in claim 1 , wherein the class retrieval circuitry is arranged to retrieve in a serial sequence the class indications for each feature data value received in the feature dataset.

8. The apparatus as claimed in claim 1 , wherein the class indications are weighted.

9. The apparatus as claimed in claim 8 , wherein weights for the class indications are determined in a training phase using feature training datasets.

10. The apparatus as claimed in claim 8 ,
wherein the classification output circuitry is responsive to reception of the class indications from the class retrieval circuitry to determine the classification by a vote amongst the class indications,
wherein weights of the class indications are used as a tiebreaker when the vote selects more than one class indication.

11. The apparatus as claimed in claim 1 , wherein each feature data value is represented by a set of fewer than 10 bits, preferably by a set of 5 bits.

12. The apparatus as claimed in claim 1 , wherein the class indications are stored in the class indications storage using a representation of fewer than 5 bits, preferably using a representation of 3 bits.

13. The apparatus as claimed in claim 1 , wherein the apparatus is a wearable device.

14. The apparatus as claimed in claim 1 , wherein the feature dataset input is coupled to a plurality of sensors each providing a respective feature data value.

15. The apparatus as claimed in claim 1 , wherein the apparatus is embodied comprising fewer than 1000 logic gates and preferably comprising fewer than 500 logic gates.

16. A method of operating an apparatus embodied as a plastic fabricated device, the method comprising:
receiving by feature dataset input circuitry included in the plastic fabricated device at a feature dataset input a feature dataset comprising multiple feature data values indicative of a set of features, wherein each feature data value is represented by a set of bits;
retrieving by class retrieval circuitry included in the plastic fabricated device from class indications storage a class indication for each feature data value received in the feature dataset, wherein the class indications are predetermined and stored in the class indications storage for each possible value of the set of bits for each feature, and wherein the class indications storage has a look-up table format and the retrieving, by class retrieval circuitry, comprises performing a look-up procedure with respect to the look-up table format for each feature data value received in the feature dataset; and
determining by classification output circuitry included in the plastic fabricated device a classification by a vote amongst the class indications.

17. Apparatus embodied as a plastic fabricated device, the apparatus comprising:
means, included as part of the plastic fabricated device, for receiving a feature dataset comprising multiple feature data values indicative of a set of features, wherein each feature data value is represented by a set of bits;
means, included as part of the plastic fabricated device, for retrieving from means for storing class indications a class indication for each feature data value received in the feature dataset, wherein class indications are predetermined and stored in the means for storing class indications for each possible value of the set of bits for each feature, wherein the means for storing class indications has a look-up table format, and wherein the means for retrieving is arranged to perform a look-up procedure with respect to the look-up table format for each feature data value received in the feature dataset; and
means for determining a classification by a vote amongst the class indications.

Assignees

• Advanced Risc Mach Ltd

Inventors

• Özer Emre
• Brown Gavin
• Reynolds Charles Edward Michael
• Kufel Jedrzej
• Biggs John Philip

Classifications

• G16H40/60

  for the operation of medical equipment or devices · CPC title

• A61B5/6813

  Specially adapted to be attached to a specific body part (A61B5/6802 takes precedence) · CPC title

• G16H10/65

  stored on portable record carriers, e.g. on smartcards, RFID tags or CD · CPC title

• A63B2220/836

  Sensors arranged on the body of the user · CPC title

• A61B5/6801

  specially adapted to be attached to or worn on the body surface · CPC title