18061216. SYSTEM AND METHOD FOR CONTINUAL REFINABLE NETWORK simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

SYSTEM AND METHOD FOR CONTINUAL REFINABLE NETWORK

Organization Name

SAMSUNG ELECTRONICS CO., LTD.

Inventor(s)

Sima Behpour of Sunnyvale CA (US)

Yilin Shen of Santa Clara CA (US)

Hongxia Jin of San Jose CA (US)

SYSTEM AND METHOD FOR CONTINUAL REFINABLE NETWORK - A simplified explanation of the abstract

This abstract first appeared for US patent application 18061216 titled 'SYSTEM AND METHOD FOR CONTINUAL REFINABLE NETWORK

Simplified Explanation

The abstract describes a method involving the use of a machine learning model trained with a dynamic learning rate and directed gradients to flat local minima. The method includes accessing the machine learning model, receiving an input, providing the input to the model, receiving an output, and instructing actions based on the output.

• The method involves accessing and using a machine learning model on an electronic device.
• The machine learning model is trained using directed gradients to flat local minima.
• A dynamic learning rate is used for one or more additional tasks during training.
• The method receives an input from an input source.
• The input is provided to the machine learning model.
• The method receives an output from the machine learning model.
• Based on the output, the method instructs at least one action.

Potential Applications

• This method can be applied in various fields where machine learning models are used, such as image recognition, natural language processing, and recommendation systems.
• It can be used in autonomous vehicles for decision-making based on sensor inputs.
• The method can be utilized in healthcare for diagnosis and treatment recommendation systems.

Problems Solved

• The method addresses the problem of training machine learning models to effectively navigate flat local minima during the learning process.
• It solves the challenge of adapting the learning rate dynamically for different tasks during training.
• The method helps in generating accurate and reliable outputs from the machine learning model.

Benefits

• By training the machine learning model with directed gradients to flat local minima, the method improves the model's ability to avoid getting stuck in suboptimal solutions.
• The use of a dynamic learning rate for additional tasks enhances the model's adaptability and performance.
• The method provides a reliable and efficient way to process inputs and generate instructive outputs based on the machine learning model's predictions.

Original Abstract Submitted

A method includes accessing, using at least one processor of an electronic device, a machine learning model. The machine learning model is trained by directing a gradient direction of gradients to one or more flat local minima and using a dynamic learning rate for one or more additional tasks. The method also includes receiving, using the at least one processor, an input from an input source. The method further includes providing, using the at least one processor, the input to the machine learning model. The method also includes receiving, using the at least one processor, an output from the machine learning model. In addition, the method includes instructing, using the at least one processor, at least one action based on the output from the machine learning model.