17948074. MACHINE TEACHING WITH METHOD OF MOMENTS simplified abstract (MICROSOFT TECHNOLOGY LICENSING, LLC)

MACHINE TEACHING WITH METHOD OF MOMENTS

Organization Name

MICROSOFT TECHNOLOGY LICENSING, LLC

Inventor(s)

Kingsuk Maitra of Fremont CA (US)

Brendan Lee Bryant of Kirkland WA (US)

Kence Anderson of Berkeley CA (US)

MACHINE TEACHING WITH METHOD OF MOMENTS - A simplified explanation of the abstract

This abstract first appeared for US patent application 17948074 titled 'MACHINE TEACHING WITH METHOD OF MOMENTS

Simplified Explanation

The techniques disclosed in this patent application involve training a machine learning model with states collected from a mechanical system, augmenting these states with fluctuating delta values derived from fixed setpoint values, and using a method of moments computation to convert delta values back into absolute values for computing a regression equation that guides the mechanical system in setting setpoints.

• Machine learning model trained with states from mechanical system
• States augmented with fluctuating delta values derived from fixed setpoint values
• Method of moments computation used to convert delta values back into absolute values
• Regression equation computed for setting setpoints in mechanical system

Potential Applications

This technology could be applied in various industries where precise control of mechanical systems is required, such as HVAC systems, manufacturing processes, and robotics.

Problems Solved

1. Limited exploration of possible setpoint values due to states with little to no variation 2. Difficulty in training machine learning models to control mechanical systems effectively

Benefits

1. Enables thorough exploration of setpoint values for better control of mechanical systems 2. Improves the efficiency and accuracy of machine learning models in controlling mechanical systems

Potential Commercial Applications

Optimizing energy consumption in HVAC systems, enhancing production processes in manufacturing, and improving the performance of robotic systems in various industries.

Possible Prior Art

One possible prior art could be the use of machine learning models to control mechanical systems, but the specific technique of augmenting states with fluctuating delta values derived from fixed setpoint values may be unique to this patent application.

How does this technology compare to traditional control methods for mechanical systems?

This article does not provide a direct comparison between this technology and traditional control methods for mechanical systems. It would be interesting to see a study or analysis that evaluates the effectiveness and efficiency of this innovation compared to conventional control techniques.

What are the potential limitations or challenges of implementing this technology in real-world applications?

The article does not address the potential limitations or challenges that may arise when implementing this technology in practical settings. It would be beneficial to understand any constraints or obstacles that could impact the adoption and deployment of this innovation.

Original Abstract Submitted

The techniques disclosed herein enable utilizing a full range of setpoint values to control a mechanical system. A machine learning model is trained with states collected from the mechanical system. Some of the states may have little to no variation, limiting exploration of possible setpoint values when training the model. To enable a more thorough exploration of possible setpoint values, the states are augmented with a fluctuating delta value that is derived from a fixed setpoint value. For example, a delta outside air temperature may be computed by subtracting outside air temperature, which fluctuates, from a fixed chilled water setpoint. A method of moments computation converts delta values inferred by the model back into absolute values. The absolute values are used to compute a regression equation that is usable by the mechanical system to compute a setpoint action for a given set of input states.