公开(公告)号：US12231490B2

公开(公告)日：2025-02-18

申请号：US18550856

申请日：2022-06-09

Applicant: Intel Corporation

Inventor： Mustafa Riza Akdeniz ,  Arjun Anand ,  Ravikumar Balakrishnan ,  Sagar Dhakal ,  Nageen Himayat

IPC: G06F13/00 ,  G06N3/098 ,  H04L67/10 ,  H04L67/289

Abstract: An apparatus of an edge computing node, a method, and a machine-readable storage medium. The apparatus is to decode messages from a plurality of clients within the edge computing network, the messages including respective coded data for respective ones of the plurality of clients; computing estimates of metrics related to a global model for federated learning using the coded data, the metrics including a gradient on the coded data; use the metrics to update the global model to generate an updated global model, wherein the edge computing node is to update the global model by calculating the gradient on the coded data based on a linear fit of the global model to estimated labels from the federated learning; and send a message including the updated global model for transmission to at least some of the clients.

公开(公告)号：US11943022B2

公开(公告)日：2024-03-26

申请号：US17435587

申请日：2019-03-29

Applicant: Intel Corporation

Inventor： Venkatesan Nallampatti Ekambaram ,  Yang-Seok Choi ,  Junyoung Nam ,  Feng Xue ,  Shu-ping Yeh ,  Hosein Nikopour ,  Shilpa Talwar ,  Jan Schreck ,  Nageen Himayat ,  Sagar Dhakal

IPC: H04W16/28 ,  H04B7/06 ,  H04B7/08 ,  H04W24/10 ,  H04W56/00 ,  H04W64/00 ,  H04W84/06

CPC classification number: H04B7/0617 ,  H04B7/086 ,  H04W16/28 ,  H04W24/10 ,  H04W56/001 ,  H04W84/06

Abstract: Systems and methods of beamforming and improving mmWave communications for drones are described. Multiple RF chains are used to adapt the main beam to track changes without the use of pilot signals. To reduce interference, interfering signal power is eliminated by optimizing a non-Gaussian measure to extract the interferers. The AoA of signals from a target drone on neighbouring drones and location of the neighbouring drones and base stations are used to independently corroborate the location reported by the target drone. The base station provides additional synchronization signals below 6 GHz and restricts the search/measurement space in the vertical direction. The inherent sparse structure above 6 GHz is exploited by applying different beamformers on a sounding signal and estimating the AoA and impulse response. Variations of fully digital and hybrid beamforming architectures for multi-cell DL sync and CRS measurement are described.

公开(公告)号：US20230177349A1

公开(公告)日：2023-06-08

申请号：US17920839

申请日：2021-05-29

Applicant: Intel Corporation

Inventor： Ravikumar Balakrishnan ,  Nageen Himayat ,  Mustafa Riza Akdeniz ,  Sagar Dhakal ,  Arjun Anand ,  Hesham Mostafa

IPC: G06N3/098 ,  H04L67/10

CPC classification number: G06N3/098 ,  H04L67/10

Abstract: The apparatus of an edge computing node, a system, a method and a machine-readable medium. The apparatus includes a processor to cause an initial set of weights for a global machine learning (ML) model to be transmitted a set of client compute nodes of the edge computing network; process Hessians computed by each of the client compute nodes based on a dataset stored on the client compute node; evaluate a gradient expression for the ML model based on a second dataset and an updated set of weights received from the client compute nodes; and generate a meta-updated set of weights for the global model based on the initial set of weights, the Hessians received, and the evaluated gradient expression.

公开(公告)号：US12061957B2

公开(公告)日：2024-08-13

申请号：US17665025

申请日：2022-02-04

Applicant: Intel Corporation

Inventor： Saurav Prakash ,  Sagar Dhakal ,  Yair Yona ,  Nageen Himayat ,  Shilpa Talwar

IPC: G06N20/00 ,  G06F9/50 ,  G06F18/214 ,  G06N3/08 ,  G06N7/01 ,  G06N7/08 ,  G06V10/94 ,  H04L43/08 ,  H04L41/14

CPC classification number: G06N20/00 ,  G06F9/5072 ,  G06F18/2148 ,  G06N3/08 ,  H04L43/08 ,  G06N7/01 ,  G06N7/08 ,  H04L41/145

Abstract: Systems, apparatuses, methods, and computer-readable media, are provided for distributed machine learning (ML) training using heterogeneous compute nodes in a heterogeneous computing environment, where the heterogeneous compute nodes are connected to a master node via respective wireless links. ML computations are performed by individual heterogeneous compute nodes on respective training datasets, and a master combines the outputs of the ML computations obtained from individual heterogeneous compute nodes. The ML computations are balanced across the heterogeneous compute nodes based on knowledge of network conditions and operational constraints experienced by the heterogeneous compute nodes. Other embodiments may be described and/or claimed.

公开(公告)号：US20240155025A1

公开(公告)日：2024-05-09

申请号：US18550856

申请日：2022-06-09

Applicant: Intel Corporation

Inventor： Mustafa Riza Akdeniz ,  Arjun Anand ,  Ravikumar Balakrishnan ,  Sagar Dhakal ,  Nageen Himayat

IPC: H04L67/10 ,  G06F17/18 ,  H04L67/289

CPC classification number: H04L67/10 ,  G06F17/18 ,  H04L67/289

Abstract: An apparatus of an edge computing node, a method, and a machine-readable storage medium. The apparatus is to decode messages from a plurality of clients within the edge computing network, the messages including respective coded data for respective ones of the plurality of clients; computing estimates of metrics related to a global model for federated learning using the coded data, the metrics including a gradient on the coded data; use the metrics to update the global model to generate an updated global model, wherein the edge computing node is to update the global model by calculating the gradient on the coded data based on a linear fit of the global model to estimated labels from the federated learning; and send a message including the updated global model for transmission to at least some of the clients.

公开(公告)号：US20220377614A1

公开(公告)日：2022-11-24

申请号：US17712050

申请日：2022-04-01

Applicant: Intel Corporation

Inventor： Ravikumar Balakrishnan ,  Nageen Himayat ,  Arjun Anand ,  Mustafa Riza Akdeniz ,  Sagar Dhakal ,  Mark R. Eisen ,  Navid Naderializadeh

IPC: H04W28/08 ,  H04W28/02

Abstract: An apparatus of a transmitter computing node n (TX node n) of a wireless network, one or more computer readable media, a system, and a method. The apparatus includes one or more processors to: implement machine learning (ML) based training rounds, each training round including: determining a local action value function Qn(hn, an; θn) corresponding to a value of performing a radio resource management (RRM) action an at a receiving computing node n (RX node n) associated with TX node n using policy parameter θn and based on hn, hn including channel state information at RX node n; and determining, based on an overall action value function Qtot at time t, an estimated gradient of an overall loss at time t for overall policy parameter θt(∇Lt(θt)), wherein Qtot corresponds to a mixing of local action value functions Qi(hi, ai; θi) for all TX nodes i in the network at time t including TX node n; and determine, in response to a determination that ∇Lt(θt) is close to zero for various values of t during training, a trained local action value function Qn,trained to generate a trained action value relating to data communication between TX node n and RX node n.

公开(公告)号：US11423254B2

公开(公告)日：2022-08-23

申请号：US16368716

申请日：2019-03-28

Applicant: Intel Corporation

Inventor： Saurav Prakash ,  Sagar Dhakal ,  Yair Yona ,  Nageen Himayat ,  Shilpa Talwar

IPC: G06F9/50 ,  G06N20/00 ,  G06N3/08 ,  G06F15/16 ,  G06K9/62 ,  G06V10/94

Abstract: Systems, apparatuses, methods, and computer-readable media are provided for load partitioning in distributed machine learning (ML) training using heterogeneous compute nodes in a heterogeneous computing environment, where the heterogeneous compute nodes are connected to a master node via respective wireless links. ML computations are performed by individual heterogeneous compute nodes on respective load partitions. The ML computations are balanced across the heterogeneous compute nodes based on knowledge of respective computational and link parameters of the heterogeneous compute nodes. Other embodiments may be described and/or claimed.

公开(公告)号：US20220237515A1

公开(公告)日：2022-07-28

申请号：US17665025

申请日：2022-02-04

Applicant: Intel Corporation

Inventor： Saurav Prakash ,  Sagar Dhakal ,  Yair Yona ,  Nageen Himayat ,  Shilpa Talwar

IPC: G06N20/00 ,  H04L41/14 ,  G06K9/62 ,  G06F9/50 ,  G06N3/08 ,  H04L43/16 ,  H04L43/08 ,  G06V10/94 ,  G06V10/96

Abstract: Systems, apparatuses, methods, and computer-readable media, are provided for distributed machine learning (ML) training using heterogeneous compute nodes in a heterogeneous computing environment, where the heterogeneous compute nodes are connected to a master node via respective wireless links. ML computations are performed by individual heterogeneous compute nodes on respective training datasets, and a master combines the outputs of the ML computations obtained from individual heterogeneous compute nodes. The ML computations are balanced across the heterogeneous compute nodes based on knowledge of network conditions and operational constraints experienced by the heterogeneous compute nodes. Other embodiments may be described and/or claimed.

公开(公告)号：US20220158702A1

公开(公告)日：2022-05-19

申请号：US17435587

申请日：2019-03-29

Applicant: Intel Corporation

Inventor： Venkatesan Nallampatti Ekambaram ,  Yang-Seok Choi ,  Junyoung Nam ,  Feng Xue ,  Shu-ping Yeh ,  Hosein Nikopour ,  Shilpa Talwar ,  Jan Schreck ,  Nageen Himayat ,  Sagar Dhakal

IPC: H04B7/06 ,  H04B7/08 ,  H04W16/28 ,  H04W24/10 ,  H04W56/00

Abstract: Systems and methods of beamforming and improving mmWave communications for drones are described. Multiple RF chains are used to adapt the main beam to track changes without the use of pilot signals. To reduce interference, interfering signal power is eliminated by optimizing a non-Gaussian measure to extract the interferers. The AoA of signals from a target drone on neighbouring drones and location of the neighbouring drones and base stations are used to independently corroborate the location reported by the target drone. The base station provides additional synchronization signals below 6 GHz and restricts the search/measurement space in the vertical direction. The inherent sparse structure above 6 GHz is exploited by applying different beamformers on a sounding signal and estimating the AoA and impulse response. Variations of fully digital and hybrid beamforming architectures for multi-cell DL sync and CRS measurement are described.

公开(公告)号：US20230189319A1

公开(公告)日：2023-06-15

申请号：US17921549

申请日：2021-06-26

Applicant: Intel Corporation

Inventor： Mustafa Riza Akdeniz ,  Nageen Himayat ,  Ravikumar Balakrishnan ,  Sagar Dhakal ,  Mark R. Eisen ,  Navid Naderializadeh

IPC: H04W72/542 ,  H04W24/02 ,  G06N3/08

CPC classification number: H04W72/542 ,  H04W24/02 ,  G06N3/08

Abstract: In one embodiment, a machine learning (ML) model for determining radio resource management (RRM) decisions is updated, with ML model parameters being shared between RRM decision makers to update the model. The updates may include local operations (between an AP and UE pair) to update local primal and dual parameters of the ML model, and global operations (between other devices in the network) to exchange/update global parameters of the ML model.