View source for 17902626. METHOD OF LOAD FORECASTING VIA KNOWLEDGE DISTILLATION, AND AN APPARATUS FOR THE SAME simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

=METHOD OF LOAD FORECASTING VIA KNOWLEDGE DISTILLATION, AND AN APPARATUS FOR THE SAME=


==Organization Name==

[[:Category:SAMSUNG ELECTRONICS CO., LTD.|SAMSUNG ELECTRONICS CO., LTD.]]


[[Category:SAMSUNG ELECTRONICS CO., LTD.]]

==Inventor(s)==

[[:Category:Chengming Hu of Montreal (CA)|Chengming Hu of Montreal (CA)]][[Category:Chengming Hu of Montreal (CA)]]

[[:Category:Xi Chen of Montreal (CA)|Xi Chen of Montreal (CA)]][[Category:Xi Chen of Montreal (CA)]]

[[:Category:Amal Feriani of Winnipeg (CA)|Amal Feriani of Winnipeg (CA)]][[Category:Amal Feriani of Winnipeg (CA)]]

[[:Category:Ju Wang of BRISSARD (CA)|Ju Wang of BRISSARD (CA)]][[Category:Ju Wang of BRISSARD (CA)]]

[[:Category:Jikun Kang of Montreal (CA)|Jikun Kang of Montreal (CA)]][[Category:Jikun Kang of Montreal (CA)]]

[[:Category:Xue Liu of Montreal (CA)|Xue Liu of Montreal (CA)]][[Category:Xue Liu of Montreal (CA)]]

[[:Category:Gregory Lewis Dudek of Westmount CA (US)|Gregory Lewis Dudek of Westmount CA (US)]][[Category:Gregory Lewis Dudek of Westmount CA (US)]]

[[:Category:Seowoo Jang of Seoul (KR)|Seowoo Jang of Seoul (KR)]][[Category:Seowoo Jang of Seoul (KR)]]

==METHOD OF LOAD FORECASTING VIA KNOWLEDGE DISTILLATION, AND AN APPARATUS FOR THE SAME - A simplified explanation of the abstract==


This abstract first appeared for US patent application 17902626 titled 'METHOD OF LOAD FORECASTING VIA KNOWLEDGE DISTILLATION, AND AN APPARATUS FOR THE SAME

==Simplified Explanation==

The patent application describes a system that uses artificial intelligence (AI) models to predict communication traffic load in a target base station. Here are the key points:

* The system obtains AI models from source base stations, which act as "teachers" for the prediction task.
* It also collects target traffic data from the target base station, which is the base station for which the traffic load needs to be predicted.
* The system integrates the predictions of the teacher AI models based on their importance weights to obtain an overall prediction.
* A student AI model is trained to minimize the difference between its own prediction and the integrated teacher prediction on the target traffic data.
* The importance weights of the teacher AI models are updated to minimize the difference between the student's prediction and the integrated teacher prediction.
* The student AI model is then updated based on the updated importance weights, and it is used to predict the communication traffic load of the target base station.

Potential applications of this technology:

* Telecommunication companies can use this system to accurately predict the communication traffic load in their base stations, helping them optimize network resources and improve service quality.
* The system can be used in smart cities to monitor and manage the traffic load in different areas, enabling better traffic flow management and reducing congestion.

Problems solved by this technology:

* Traditional methods of predicting communication traffic load may be inaccurate and inefficient.
* This system provides a more accurate and efficient way of predicting traffic load by integrating multiple AI models and updating them based on the student model's performance.

Benefits of this technology:

* Improved accuracy: By integrating multiple AI models and updating them based on the student model's performance, the system can provide more accurate predictions of communication traffic load.
* Efficient resource allocation: Accurate predictions help telecommunication companies optimize their network resources, leading to better service quality and reduced costs.
* Traffic management: In smart cities, accurate predictions can help authorities manage traffic flow more effectively, reducing congestion and improving overall transportation efficiency.


==Original Abstract Submitted==

A server may obtain teacher artificial intelligence (AI) models from source base stations; obtain target traffic data from a target base station; obtain an integrated teacher prediction based on the target traffic data by integrating teacher prediction results of the teacher AI models based on teacher importance weights; obtain a student AI model that is trained to converge a student loss on the target traffic data; update the teacher importance weights to converge a teacher loss between a student prediction of the student AI model on the target traffic data, and the integrated teacher prediction of the teacher AI models on the target traffic data; update the student AI model based on the updated teacher importance weights being applied to the teacher prediction results of the teacher AI models; and predict a communication traffic load of the target base station using the updated student AI model.



[[Category:H04W28/16]]
[[Category:G06N20/00]]
[[Category:H04W24/02]]
[[Category:H04L41/16]]