18115520. TECHNIQUES FOR COMPRESSING NEURAL NETWORKS simplified abstract (NVIDIA Corporation)

From WikiPatents
Jump to navigation Jump to search

TECHNIQUES FOR COMPRESSING NEURAL NETWORKS

Organization Name

NVIDIA Corporation

Inventor(s)

Chong Yu of Shanghai (CN)

TECHNIQUES FOR COMPRESSING NEURAL NETWORKS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18115520 titled 'TECHNIQUES FOR COMPRESSING NEURAL NETWORKS

Simplified Explanation

The patent application describes apparatuses, systems, and techniques for compressing neural networks based on accuracy and performance.

  • One or more first neural networks are used to select compressed neural networks.
  • Selection is based on accuracy and performance of the compressed neural networks.

Potential Applications

The technology could be applied in various fields such as:

  • Image recognition
  • Natural language processing
  • Autonomous vehicles

Problems Solved

  • Reducing the size of neural networks
  • Improving efficiency and speed of neural network operations

Benefits

  • Faster processing speeds
  • Reduced memory usage
  • Improved overall performance of neural networks

Potential Commercial Applications

The technology could be utilized in industries such as:

  • Healthcare for medical image analysis
  • Finance for fraud detection
  • Manufacturing for quality control

Possible Prior Art

One example of prior art in this field is the use of pruning techniques to reduce the size of neural networks while maintaining performance levels.

Unanswered Questions

How does the technology handle complex neural network architectures?

The article does not provide specific details on how the technology deals with intricate neural network structures.

What impact does network compression have on training times?

The article does not address the potential effects of network compression on the training duration of neural networks.


Original Abstract Submitted

Apparatuses, systems, and techniques to compress neural networks. In at least one embodiment, one or more first neural networks are used to cause one or more compressed neural networks to be selected based, at least in part, on accuracy and performance of the one or more compressed neural networks.