18322988. EFFICIENT CONVOLUTION IN MACHINE LEARNING ENVIRONMENTS simplified abstract (Intel Corporation)

From WikiPatents
Jump to navigation Jump to search

EFFICIENT CONVOLUTION IN MACHINE LEARNING ENVIRONMENTS

Organization Name

Intel Corporation

Inventor(s)

Dhawal Srivastava of Scottsdale AZ (US)

EFFICIENT CONVOLUTION IN MACHINE LEARNING ENVIRONMENTS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18322988 titled 'EFFICIENT CONVOLUTION IN MACHINE LEARNING ENVIRONMENTS

Simplified Explanation

The abstract describes a mechanism for facilitating smart convolution in machine learning environments. The apparatus includes processors, detection and selection logic, and filter generation and storage logic.

  • The apparatus includes one or more processors, including graphics processors, for performing smart convolution in machine learning environments.
  • The detection and selection logic is responsible for identifying input images that have geometric shapes associated with a specific object for training a neural network.
  • The filter generation and storage logic generates weights for filters based on the geometric shapes identified in the input images.
  • The filter logic further organizes the filters into filter groups based on common geometric shapes and stores them in bins corresponding to each geometric shape.

Potential applications of this technology:

  • This mechanism can be used in various machine learning applications that involve image recognition and object detection.
  • It can enhance the training process of neural networks by efficiently selecting and generating filters based on the geometric shapes present in the input images.

Problems solved by this technology:

  • Traditional convolution methods in machine learning may not effectively handle input images with complex geometric shapes.
  • This mechanism solves the problem of efficiently selecting and generating filters based on the geometric shapes associated with the objects in the input images.

Benefits of this technology:

  • The smart convolution mechanism improves the accuracy and efficiency of machine learning models by focusing on relevant geometric shapes in the input images.
  • It reduces the computational complexity of the training process by organizing filters into groups and bins based on common geometric shapes.


Original Abstract Submitted

A mechanism is described for facilitating smart convolution in machine learning environments. An apparatus of embodiments, as described herein, includes one or more processors including one or more graphics processors, and detection and selection logic to detect and select input images having a plurality of geometric shapes associated with an object for which a neural network is to be trained. The apparatus further includes filter generation and storage logic (“filter logic”) to generate weights providing filters based on the plurality of geometric shapes, where the filter logic is further to sort the filters in filter groups based on common geometric shapes of the plurality of geographic shapes, and where the filter logic is further to store the filter groups in bins based on the common geometric shapes, wherein each bin corresponds to a geometric shape.

Retrieved from "http://wikipatents.org/index.php?title=18322988._EFFICIENT_CONVOLUTION_IN_MACHINE_LEARNING_ENVIRONMENTS_simplified_abstract_(Intel_Corporation)&oldid=16642"