Intel corporation (20240256825). CONVOLUTIONAL NEURAL NETWORK OPTIMIZATION MECHANISM simplified abstract

CONVOLUTIONAL NEURAL NETWORK OPTIMIZATION MECHANISM

Organization Name

intel corporation

Inventor(s)

Liwei Ma of Beijing (CN)

Elmoustapha Ould-ahmed-vall of Chandler AZ (US)

Barath Lakshmanan of Chandler AZ (US)

Ben J. Ashbaugh of Folsom CA (US)

Jingyi Jin of Folsom CA (US)

Jeremy Bottleson of Rancho Cordova CA (US)

Mike B. Macpherson of Portland OR (US)

Kevin Nealis of San Jose CA (US)

Dhawal Srivastava of Phoenix AZ (US)

Joydeep Ray of Folsom CA (US)

Ping T. Tang of Edison NJ (US)

Michael S. Strickland of Sunnyvale CA (US)

Xiaoming Chen of Shanghai (CN)

Anbang Yao of Beijing (CN)

Tatiana Shpeisman of Menlo Park CA (US)

Altug Koker of El Dorado Hills CA (US)

Abhishek R. Appu of El Dorado Hills CA (US)

CONVOLUTIONAL NEURAL NETWORK OPTIMIZATION MECHANISM - A simplified explanation of the abstract

This abstract first appeared for US patent application 20240256825 titled 'CONVOLUTIONAL NEURAL NETWORK OPTIMIZATION MECHANISM

Simplified Explanation

A library of machine learning primitives is used to optimize a machine learning model for more efficient inference operations. In one example, a trained convolutional neural network (CNN) model is improved through pruning, convolution window optimization, and quantization.

• Trained CNN model optimization
• Pruning, convolution window optimization, and quantization techniques used
• Aim to enhance efficiency of inference operations

Key Features and Innovation

• Library of machine learning primitives for model optimization
• Utilization of pruning, convolution window optimization, and quantization for trained CNN models
• Focus on improving efficiency of inference operations

Potential Applications

This technology can be applied in various fields such as image recognition, natural language processing, and autonomous vehicles.

Problems Solved

• Enhances the efficiency of inference operations
• Improves the performance of trained CNN models
• Reduces computational resources required for machine learning tasks

Benefits

• Faster inference operations
• Reduced computational costs
• Improved accuracy of machine learning models

Commercial Applications

Optimizing machine learning models using these techniques can benefit industries such as healthcare, finance, and e-commerce by improving the efficiency and accuracy of their AI systems.

Questions about Machine Learning Model Optimization

1. How does pruning contribute to optimizing a trained CNN model?

  Pruning helps remove unnecessary connections in the model, reducing its size and computational requirements.

2. What are the advantages of quantization in machine learning model optimization?

  Quantization reduces the precision of numerical values in the model, leading to smaller memory footprint and faster computations.

Original Abstract Submitted

a library of machine learning primitives is provided to optimize a machine learning model to improve the efficiency of inference operations. in one embodiment a trained convolutional neural network (cnn) model is processed into a trained cnn model via pruning, convolution window optimization, and quantization.