View source for 17987312. METHOD AND DEVICE WITH NEURAL NETWORK TRAINING AND IMAGE PROCESSING simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)

=METHOD AND DEVICE WITH NEURAL NETWORK TRAINING AND IMAGE PROCESSING=


==Organization Name==

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


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

==Inventor(s)==

[[:Category:Xiaolei Zhang of Xi'an (CN)|Xiaolei Zhang of Xi'an (CN)]][[Category:Xiaolei Zhang of Xi'an (CN)]]

[[:Category:Zhaohui Lv of Xi'an (CN)|Zhaohui Lv of Xi'an (CN)]][[Category:Zhaohui Lv of Xi'an (CN)]]

[[:Category:Mingming Fan of Xi'an (CN)|Mingming Fan of Xi'an (CN)]][[Category:Mingming Fan of Xi'an (CN)]]

[[:Category:Zixuan Leng of Xi'an (CN)|Zixuan Leng of Xi'an (CN)]][[Category:Zixuan Leng of Xi'an (CN)]]

==METHOD AND DEVICE WITH NEURAL NETWORK TRAINING AND IMAGE PROCESSING - A simplified explanation of the abstract==


This abstract first appeared for US patent application 17987312 titled 'METHOD AND DEVICE WITH NEURAL NETWORK TRAINING AND IMAGE PROCESSING

==Simplified Explanation==

The abstract describes a method implemented by a processor for training a student network using a teacher network. The method involves generating outputs from multiple layers of the teacher network based on an input image, generating pseudo labels based on these outputs, generating an output from the student network based on the same input image, generating prediction results from the teacher network based on the student network's output, and updating the student network using the pseudo labels and prediction results.

* The method involves using a teacher network and a student network for training.
* The teacher network generates outputs from multiple layers based on an input image.
* Pseudo labels are generated based on the outputs of the teacher network.
* The student network generates an output based on the same input image.
* Prediction results are generated from the teacher network based on the student network's output.
* The student network is updated using the pseudo labels and prediction results.

== Potential Applications ==
* This method can be applied in various fields that require training neural networks, such as computer vision, natural language processing, and speech recognition.
* It can be used to improve the performance and accuracy of the student network by leveraging the knowledge and outputs of the teacher network.

== Problems Solved ==
* This method addresses the problem of training a student network without the need for extensive labeled training data.
* It solves the problem of improving the performance of the student network by utilizing the knowledge and outputs of a teacher network.

== Benefits ==
* By using the teacher network's outputs and pseudo labels, the student network can be trained more effectively and efficiently.
* This method reduces the reliance on labeled training data, making it easier and more cost-effective to train neural networks.
* It allows for knowledge transfer from the teacher network to the student network, leading to improved performance and accuracy.


==Original Abstract Submitted==

A processor-implemented method includes: generating a first output of each of two or more layers of a teacher network, based on a first image; generating pseudo labels respectively corresponding to the first outputs, based on the first outputs; generating a second output using one or more layers of a student network comprising an output layer, based on the first image; generating prediction results respectively corresponding to the two or more layers of the teacher network, based on the second output; and training the student network by updating the student network based on the pseudo labels and the prediction results.



[[Category:G06V10/82]]
[[Category:G06N3/04]]