Search

US-20260129429-A1 - MODEL TUNING FOR CROSS NODE MACHINE LEARNING

US20260129429A1US 20260129429 A1US20260129429 A1US 20260129429A1US-20260129429-A1

Abstract

Methods, systems, and devices for wireless communications are described. A user equipment (UE) may obtain data samples for a first machine learning model associated with a task at the UE. The first set of parameters may be associated with the first machine learning model. The UE may transmit a capability message indicating a capability of the UE to perform a tuning procedure of the first machine learning model, and the UE may perform the tuning procedure of the first machine learning model to obtain a second set of parameters associated with the first machine learning model based on the capability of the UE to perform the tuning procedure of the first machine learning model. The capability of the UE to perform the tuning procedure may be one of an online tuning capability or an offline tuning capability.

Inventors

  • Bongyong SONG
  • Taesang Yoo
  • Chenxi HAO
  • Jay Kumar Sundararajan
  • June Namgoong

Assignees

  • QUALCOMM INCORPORATED

Dates

Publication Date
20260507
Application Date
20221122

Claims (20)

  1. 1 . A method for wireless communication at a user equipment (UE), comprising: obtaining data samples for a first machine learning model associated with a task at the UE, wherein a first set of parameters is associated with the first machine learning model; transmitting a capability message indicating a capability of the UE to perform a tuning procedure of the first machine learning model; performing the tuning procedure of the first machine learning model to obtain a second set of parameters associated with the first machine learning model based at least in part on the capability of the UE to perform the tuning procedure of the first machine learning model; and transmitting, to a network entity, a message indicating at least a portion of the second set of parameters based at least in part on performing the tuning procedure of the first machine learning model.
  2. 2 . The method of claim 1 , wherein the first machine learning model comprises an encoder portion of a second machine learning model, and a third machine learning model comprises a decoder portion of the second machine learning model.
  3. 3 . The method of claim 2 , wherein performing the tuning procedure of the first machine learning model further comprises: receiving a set of parameters associated with a loss function.
  4. 4 . The method of claim 3 , further comprising: transmitting a message associated with a forward propagation procedure.
  5. 5 . The method of claim 3 , further comprising: receiving a message associated with a backward propagation procedure for adjusting parameters associated with an encoder, wherein the message indicates a gradient associated with the loss function; and updating the parameters associated with the encoder based at least in part on the message.
  6. 6 . The method of claim 3 , further comprising: receiving the set of parameters associated with the loss function; and updating parameters associated with the decoder portion of the second machine learning model based at least in part on the set of parameters.
  7. 7 . The method of claim 1 , wherein transmitting the capability message comprises: transmitting an indication of a set of machine learning models supported by the UE.
  8. 8 . The method of claim 1 , wherein the task comprises a Channel State Information (CSI) feedback task.
  9. 9 . The method of claim 1 , wherein performing the tuning procedure of the first machine learning model further comprises: updating parameters associated with an encoder, parameters associated with a decoder, or both using the second set of parameters.
  10. 10 . The method of claim 9 , wherein performing the tuning procedure of the first machine learning model comprises: performing a tuning procedure.
  11. 11 . (canceled)
  12. 12 . The method of claim 9 , wherein performing the tuning procedure of the first machine learning model comprises: performing an offline tuning procedure.
  13. 13 . The method of claim 12 , wherein performing the offline tuning procedure of the first machine learning model comprises: updating the second set of parameters associated with the encoder using the first set of parameters for the first machine learning model in performing the task.
  14. 14 . (canceled)
  15. 15 . The method of claim 1 , wherein performing the tuning procedure further comprises: receiving a first indication from the network entity associated with performing the tuning procedure of the first machine learning model, wherein the first indication comprises an activation status or an allowed status; and transmitting a second indication to the network entity in response to the first indication, wherein the second indication comprises an activation indication associated with starting to perform the tuning procedure or a deactivation indication associated with stopping the tuning procedure.
  16. 16 . The method of claim 15 , further comprising: transmitting an activation request to the network entity, wherein receiving the first indication is based at least in part on the activation request.
  17. 17 . The method of claim 15 , further comprising: receiving a fourth indication from the network entity, wherein the fourth indication comprises a deactivation indication associated with stopping the tuning procedure.
  18. 18 . The method of claim 1 , wherein performing the tuning procedure of the first machine learning model further comprises: updating parameters associated with an encoder, parameters associated with a decoder, or both using the second set of parameters based at least in part on a gradient associated with the first set of parameters and the second set of parameters.
  19. 19 . The method of claim 18 , wherein updating parameters associated with an encoder, parameters associated with a decoder, or both further comprises: receiving a set of parameters associated with a loss function; and transmitting a message associated with a forward propagation procedure.
  20. 20 . (canceled)

Description

CROSS REFERENCE The present application is a 371 national stage filing of International PCT Application No. PCT/CN2022/133369 by SONG et al. entitled “MODEL TUNING FOR CROSS NODE MACHINE LEARNING,” filed Nov. 22, 2022, which is assigned to the assignee hereof, and which is expressly incorporated by reference in its entirety herein. FIELD OF TECHNOLOGY The following relates to wireless communications, including model tuning for cross node machine learning. BACKGROUND Wireless communications systems are widely deployed to provide various types of communication content such as voice, video, packet data, messaging, broadcast, and so on. These systems may be capable of supporting communication with multiple users by sharing the available system resources (e.g., time, frequency, and power). Examples of such multiple-access systems include fourth generation (4G) systems such as Long Term Evolution (LTE) systems, LTE-Advanced (LTE-A) systems, or LTE-A Pro systems, and fifth generation (5G) systems which may be referred to as New Radio (NR) systems. These systems may employ technologies such as code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), or discrete Fourier transform spread orthogonal frequency division multiplexing (DFT-S-OFDM). A wireless multiple-access communications system may include one or more base stations, each supporting wireless communication for communication devices, which may be known as user equipment (UE). SUMMARY The described techniques relate to improved methods, systems, devices, and apparatuses that support model tuning for cross node machine learning. Generally, the described techniques enable a user equipment (UE) to autonomously perform a tuning (e.g., fine tuning) procedure for a machine learning model used in communications between the UE and a network entity. For example, the UE may receive data samples (e.g., a training data set) from the network entity to train the machine learning model. The UE may transmit a capability message to the network entity, and the capability message may indicate whether the UE may autonomously perform the tuning procedure. The UE may generate a second training data set based on the tuning procedure, and the UE or the network entity may use the second training data set to perform a tuning procedure of a second machine learning model. A method for wireless communication at a UE is described. The method may include obtaining data samples for a first machine learning model associated with a task at the UE, where a first set of parameters is associated with the first machine learning model, transmitting a capability message indicating a capability of the UE to perform a tuning procedure of the first machine learning model, performing the tuning procedure of the first machine learning model to obtain a second set of parameters associated with the first machine learning model based on the capability of the UE to perform the tuning procedure of the first machine learning model, and transmitting, to a network entity, a message indicating at least a portion of the second set of parameters based on performing the tuning procedure of the first machine learning model. An apparatus for wireless communication at a UE is described. The apparatus may include a processor, memory coupled with the processor, and instructions stored in the memory. The instructions may be executable by the processor to cause the apparatus to obtain data samples for a first machine learning model associated with a task at the UE, where a first set of parameters is associated with the first machine learning model, transmit a capability message indicating a capability of the UE to perform a tuning procedure of the first machine learning model, perform the tuning procedure of the first machine learning model to obtain a second set of parameters associated with the first machine learning model based on the capability of the UE to perform the tuning procedure of the first machine learning model, and transmit, to a network entity, a message indicating at least a portion of the second set of parameters based on performing the tuning procedure of the first machine learning model. Another apparatus for wireless communication at a UE is described. The apparatus may include means for obtaining data samples for a first machine learning model associated with a task at the UE, where a first set of parameters is associated with the first machine learning model, means for transmitting a capability message indicating a capability of the UE to perform a tuning procedure of the first machine learning model, means for performing the tuning procedure of the first machine learning model to obtain a second set of parameters associated with the first machine learning model based on the capability of the UE to perform the tuning procedure of the first machine learning model, and means for transmitting, to a network entity,