18237887. LSTM CIRCUIT WITH SELECTIVE INPUT COMPUTATION simplified abstract (Intel Corporation)

From WikiPatents
Jump to navigation Jump to search

LSTM CIRCUIT WITH SELECTIVE INPUT COMPUTATION

Organization Name

Intel Corporation

Inventor(s)

Ram Krishnamurthy of Portland OR (US)

Gregory K. Chen of Portland OR (US)

Raghavan Kumar of Hillsboro OR (US)

Phil Knag of Hillsboro OR (US)

Huseyin Ekin Sumbul of Portland OR (US)

LSTM CIRCUIT WITH SELECTIVE INPUT COMPUTATION - A simplified explanation of the abstract

This abstract first appeared for US patent application 18237887 titled 'LSTM CIRCUIT WITH SELECTIVE INPUT COMPUTATION

Simplified Explanation

The patent application describes an apparatus that includes a long short term memory (LSTM) circuit with a multiply accumulate circuit (MAC). The MAC circuit has a feature that allows it to use a stored product term instead of performing a multiplication operation, if the accumulation of differences between consecutive input values has not reached a certain threshold.

  • The apparatus includes a long short term memory (LSTM) circuit with a multiply accumulate circuit (MAC).
  • The MAC circuit has circuitry that can use a stored product term instead of performing a multiplication operation.
  • This feature is utilized when the accumulation of differences between consecutive input values has not reached a threshold.
  • The stored product term is relied upon to determine the product term in such cases.
  • The apparatus provides a more efficient way of handling calculations in the LSTM circuit.

Potential Applications

  • Artificial intelligence and machine learning systems
  • Natural language processing and speech recognition systems
  • Robotics and autonomous systems
  • Financial analysis and prediction models

Problems Solved

  • Reduces the computational load on the LSTM circuit by avoiding unnecessary multiplication operations.
  • Improves the efficiency and speed of calculations in the LSTM circuit.
  • Helps to optimize the performance of AI and machine learning systems.

Benefits

  • Faster and more efficient processing of data in LSTM circuits.
  • Reduced power consumption and improved energy efficiency.
  • Enhanced performance and accuracy of AI and machine learning models.
  • Cost savings in terms of hardware requirements for AI systems.


Original Abstract Submitted

An apparatus is described. The apparatus includes a long short term memory (LSTM) circuit having a multiply accumulate circuit (MAC). The MAC circuit has circuitry to rely on a stored product term rather than explicitly perform a multiplication operation to determine the product term if an accumulation of differences between consecutive, preceding input values has not reached a threshold.