17940935. Model Inversion in Integrated Circuit Devices having Analog Inference Capability simplified abstract (Micron Technology, Inc.)

From WikiPatents
Jump to navigation Jump to search

Model Inversion in Integrated Circuit Devices having Analog Inference Capability

Organization Name

Micron Technology, Inc.

Inventor(s)

Poorna Kale of Folsom CA (US)

Model Inversion in Integrated Circuit Devices having Analog Inference Capability - A simplified explanation of the abstract

This abstract first appeared for US patent application 17940935 titled 'Model Inversion in Integrated Circuit Devices having Analog Inference Capability

Simplified Explanation

The patent application describes a device with a memory cell array that is configured with inverted weight data for multiplication and accumulation operations. The memory cell array has programmable threshold voltages for each cell to perform these operations, with multiple regions operating in parallel. The device includes a logic circuit to adjust computation results for weight inversion and generate an output result based on results from different regions.

  • Memory cell array with inverted weight data
  • Programmable threshold voltages for multiplication and accumulation operations
  • Multiple regions operating in parallel
  • Logic circuit to adjust computation results for weight inversion
  • Output result generated based on results from different regions

Potential Applications

This technology could be applied in:

  • Neural network accelerators
  • Signal processing applications
  • Machine learning algorithms

Problems Solved

This technology helps in:

  • Improving efficiency in multiplication and accumulation operations
  • Reducing computational errors
  • Enhancing performance in parallel processing tasks

Benefits

The benefits of this technology include:

  • Faster computation speeds
  • Higher accuracy in results
  • Increased efficiency in memory usage

Potential Commercial Applications

Optimizing Inverted Weight Data Memory Cell Arrays for Multiplication and Accumulation Operations

Unanswered Questions

How does this technology compare to existing methods for multiplication and accumulation operations?

This technology offers improved efficiency and accuracy compared to traditional methods by utilizing inverted weight data and parallel processing regions.

What impact could this technology have on the development of artificial intelligence systems?

This technology could significantly enhance the performance of AI systems by providing faster and more accurate computation results, leading to advancements in various AI applications.


Original Abstract Submitted

A device having a memory cell array configured with inverted weight data for operations of multiplication and accumulation. Each respective memory cell in the memory cell array has a threshold voltage programmable in a first mode to perform operations of multiplication and accumulation. The memory cell array has a plurality of regions operable in parallel to perform operations of multiplication and accumulation. The plurality of regions include a first region and a second region. At least a second portion of weight bits stored in the second region is an inverted version of a first portion of weight bits stored in the first region. The device includes a logic circuit configured to adjust a computation result of multiplication and accumulation generated using the second region to account for weight inversion and generate an output result based on a plurality of results generated using the plurality of regions respectively.