公开(公告)号：US20170331585A1

公开(公告)日：2017-11-16

申请号：US15358960

申请日：2016-11-22

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Yang YANG ,  Joseph Binamira SORIAGA ,  Jing JIANG ,  Tingfang JI

IPC: H04L1/00 ,  H04L5/00 ,  H04B7/0413

CPC classification number: H03M13/4115 ,  H03M13/413 ,  H03M13/4169 ,  H03M13/6502 ,  H04B7/0413 ,  H04L1/0054

Abstract: Certain aspects of the present disclosure relate to techniques and apparatus for enhanced decoding, for example, by providing a multi-phase tail biting convolutional code (TBCC) decoding algorithm. An exemplary method generally includes obtaining, via a wireless medium, a codeword encoded with a TBCC encoding scheme, generating metrics for candidate paths through trellis stages of a decoder, propagating information from at least one of the trellis stages to a later trellis stage, while generating the metrics, selecting a set of the candidate paths based on the propagated information, and decoding the encoded codeword by evaluating the selected set of candidate paths based, at least in part, on the generated metrics. Other aspects, embodiments, and features are claimed and described.

公开(公告)号：US20240104367A1

公开(公告)日：2024-03-28

申请号：US17934098

申请日：2022-09-21

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Debasmit DAS

IPC: G06N3/08 ,  H04B17/391

CPC classification number: G06N3/08 ,  H04B17/3913

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for training a machine learning model. An example method generally includes partitioning a machine learning model into a plurality of partitions. A request to update a respective partition of the plurality of partitions in the machine learning model is transmitted to each respective participating device of a plurality of participating devices in a federated learning scheme, and the request may specify that the respective partition is to be updated based on unique data at the respective participating device. Updates to one or more partitions in the machine learning model are received from the plurality of participating devices, and the machine learning model is updated based on the received updates.

公开(公告)号：US20240095513A1

公开(公告)日：2024-03-21

申请号：US17932809

申请日：2022-09-16

Applicant: QUALCOMM Incorporated

Inventor： Jian SHEN ,  Jamie Menjay LIN

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for surrogated federated learning. A set of intermediate activations is received at a trusted server from a node device, where the node device generated the set of intermediate activations using a first set of layers of a neural network. One or more weights associated with a second set of layers of the neural network are refined using the set of intermediate activations, and one or more weight updates corresponding to the refined one or more weights are transmitted to a federated learning system.

公开(公告)号：US20240046078A1

公开(公告)日：2024-02-08

申请号：US17817552

申请日：2022-08-04

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Jian SHEN ,  Fatih Murat PORIKLI

IPC: G06N3/04

CPC classification number: G06N3/0481

Abstract: Certain aspects of the present disclosure provide techniques for desparsified convolution. An activation tensor is received, and a convolution output is generated for the activation tensor, comprising: selecting a subset of weight elements, corresponding to a set of non-zero elements in the activation tensor, from a weight tensor, and multiplying the set of non-zero elements and the set of weight elements.

公开(公告)号：US20230298142A1

公开(公告)日：2023-09-21

申请号：US17655427

申请日：2022-03-18

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Diaa H J BADAWI ,  Hong CAI ,  Fatih Murat PORIKLI

IPC: G06T5/00 ,  G06T7/194 ,  G06T5/50

CPC classification number: G06T5/003 ,  G06T5/002 ,  G06T7/194 ,  G06T5/005 ,  G06T5/50 ,  G06T2207/20081 ,  G06T2207/20084

Abstract: Certain aspects of the present disclosure provide techniques for machine learning-based deblurring. An input image is received, and a deblurred image is generated based on the input image using a neural network, comprising: generating a feature tensor by processing the input image using a first portion of the neural network, generating a motion mask by processing the feature tensor using a motion portion of the neural network, and generating the deblurred image by processing the feature tensor and the motion mask using a deblur portion of the neural network.

公开(公告)号：US20230215157A1

公开(公告)日：2023-07-06

申请号：US18150578

申请日：2023-01-05

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Fatih Murat PORIKLI

IPC: G06V10/82 ,  G06V10/77 ,  G06V10/75

CPC classification number: G06V10/82 ,  G06V10/77 ,  G06V10/75

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for efficient processing of visual content using machine learning models. An example method generally includes generating, from an input, an embedding tensor for the input. The embedding tensor for the input is projected into a reduced-dimensional space projection of the embedding tensor based on a projection matrix. An attention value for the input is derived based on the reduced-dimensional space projection of the embedding tensor and a non-linear attention function. A match, in the reduced-dimensional space, is identified between a portion of the input and a corresponding portion of a target against which the input is evaluated based on the attention value for the input. One or more actions are taken based on identifying the match.

公开(公告)号：US20220391702A1

公开(公告)日：2022-12-08

申请号：US17805021

申请日：2022-06-01

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Yash Sanjay Bhalgat ,  Edwin Chongwoo Park

IPC: G06N3/08 ,  G06N3/04 ,  G06F9/345

Abstract: Certain aspects of the present disclosure provide techniques for kernel expansion. An input data tensor is received at a first layer in a neural network, and a first convolution is performed for a first kernel, where the first kernel has a size greater than a preferred size. Performing the first convolution comprises generating a plurality of intermediate tensors by performing a plurality of intermediate convolutions using a plurality of intermediate kernels with a size of the preferred size, and accumulating the plurality of intermediate tensors to generate an output tensor for the first convolution.

公开(公告)号：US20210287095A1

公开(公告)日：2021-09-16

申请号：US16816117

申请日：2020-03-11

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Edwin Chongwoo PARK ,  Nojun KWAK

IPC: G06N3/08 ,  G06N5/04 ,  G06N20/10 ,  G06F7/523

Abstract: A method for operating a low-bitwidth neural network includes converting a first activation to a non-negative value (e.g., absolute value). The first activation has a signed value. The sign of the activation is used to select a weight value. A product of the non-negative activation and the selected weight value is computed to determine a next activation. The next activation is quantized and supplied to a subsequent layer of the low-bitwidth neural network.

公开(公告)号：US20210158145A1

公开(公告)日：2021-05-27

申请号：US16694442

申请日：2019-11-25

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Daniel Hendricus Franciscus FONTIJNE ,  Edwin Chongwoo PARK

IPC: G06N3/08 ,  G06N3/04

Abstract: Aspects described herein provide a method including: receiving input data at a machine learning model, comprising: a plurality of processing layers; a plurality of gate logics; a plurality of gates; and a fully connected layer; determining based on a plurality of gate parameters associated with the plurality of gate logics, a subset of the plurality of processing layers with which to process the input data; processing the input data with the subset of the plurality of processing layers and the fully connected layer to generate an inference; determining a prediction loss based on the inference and a training label associated with the input data; determining an energy loss based on the subset of the plurality of processing layers used to process the input data; and optimizing the machine learning model based on: the prediction loss; the energy loss; and a prior probability associated with the training label.

公开(公告)号：US20210019593A1

公开(公告)日：2021-01-21

申请号：US16932496

申请日：2020-07-17

Applicant: QUALCOMM Incorporated

Inventor： Jamie Menjay LIN ,  Jin Won LEE ,  Jilei HOU

IPC: G06N3/04 ,  G06F11/30 ,  G06K9/62 ,  G06F17/15

Abstract: Certain aspects of the present disclosure provide techniques for performing piecewise pointwise convolution, comprising: performing a first piecewise pointwise convolution on a first subset of data received via a first branch input at a piecewise pointwise convolution layer of a convolutional neural network (CNN) model; performing a second piecewise pointwise convolution on a second subset of data received via a second branch input at the piecewise pointwise convolution layer; determining a piecewise pointwise convolution output by summing a result of the first piecewise pointwise convolution and a result of the second piecewise pointwise convolution; and providing the piecewise pointwise convolution output to a second layer of the CNN model.