17450134. ATTENUATION WEIGHT TRACKING IN GRAPH NEURAL NETWORKS simplified abstract (International Business Machines Corporation)

From WikiPatents
Jump to navigation Jump to search

ATTENUATION WEIGHT TRACKING IN GRAPH NEURAL NETWORKS

Organization Name

International Business Machines Corporation

Inventor(s)

Gandhi Sivakumar of Bentleigh (AU)

Anil Manohar Omanwar of Vikas Nagar (IN)

Vijay Ekambaram of Chennai (IN)

Smitkumar Narotambhai Marvaniya of Bangalore (IN)

ATTENUATION WEIGHT TRACKING IN GRAPH NEURAL NETWORKS - A simplified explanation of the abstract

This abstract first appeared for US patent application 17450134 titled 'ATTENUATION WEIGHT TRACKING IN GRAPH NEURAL NETWORKS

Simplified Explanation

Abstract: The patent application describes a computer-based method for predicting the states of sensors in a sensor network. This is achieved by using a temporal graph neural network to analyze the weights and changes in weights over time in the sensor network. The method then uses a regression machine learning model to forecast the sensor states based on patterns identified in the weights.

  • The method uses a temporal graph neural network to analyze the weights and changes in weights over time in a sensor network.
  • The neural network is trained using the temporal graph of the sensor network.
  • The number of processor units in the network determines patterns in the weights and changes in weights.
  • A regression machine learning model is trained using the patterns of the weights.
  • The trained model is used to forecast the sensor states for sensors in the network.

Potential Applications:

  • Predictive maintenance in industrial settings by forecasting sensor states and detecting anomalies.
  • Environmental monitoring systems to predict changes in air quality, temperature, or other sensor measurements.
  • Smart home applications to anticipate and optimize energy usage based on sensor data.

Problems Solved:

  • The method solves the problem of accurately predicting sensor states in a sensor network.
  • It addresses the challenge of analyzing complex temporal relationships and patterns in the sensor data.
  • It overcomes the limitations of traditional forecasting methods that may not capture the dynamics of a sensor network.

Benefits:

  • Improved efficiency and cost savings by enabling proactive maintenance and optimization based on accurate sensor state predictions.
  • Enhanced decision-making capabilities by providing real-time insights into the sensor network.
  • Increased reliability and performance of sensor networks through better understanding and utilization of the data.


Original Abstract Submitted

A computer implemented method for forecasting sensor states in a sensor network is provided. A number of processor units identifies weights and changes in the weights over time in a temporal graph of the sensor network using a temporal graph neural network trained using the temporal graph. The number of processor units determines patterns of the weights based on the weights and the changes in the weights over time. The number of processor units trains a regression machine learning model using the patterns of the weights to forecast sensor states for sensors in the sensor network.

Retrieved from "http://wikipatents.org/index.php?title=17450134._ATTENUATION_WEIGHT_TRACKING_IN_GRAPH_NEURAL_NETWORKS_simplified_abstract_(International_Business_Machines_Corporation)&oldid=21526"