公开(公告)号：US20250077313A1

公开(公告)日：2025-03-06

申请号：US18459239

申请日：2023-08-31

Applicant: QUALCOMM Incorporated

Inventor： Simyung CHANG ,  Kyu Woong HWANG ,  Juntae LEE ,  Kyuhong SHIM ,  Jihwan BANG ,  Seunghan YANG ,  Jaeseong YOU ,  Minseop PARK ,  Christopher LOTT

IPC: G06F9/54 ,  G06F9/48

Abstract: A processor-implemented method for generating a default adapter for context switching includes analyzing a first neural network model and one or more adapters. The first neural network model is pre-trained and each of the adapters is configured with an architecture and parameters for performing a different downstream task of a set of downstream tasks. A default adapter is defined based on a capacity of the one or more adapters. The default adapter is applied to one or more layers of the first neural network model during a context switch to a replace one of the adapters for a different task. A graph corresponding to the first neural network model is unchanged.

公开(公告)号：US20220383117A1

公开(公告)日：2022-12-01

申请号：US17824832

申请日：2022-05-25

Applicant: QUALCOMM Incorporated

Inventor： Jangho KIM ,  Juntae LEE ,  Simyung CHANG

IPC: G06N3/08 ,  G06N5/04

Abstract: A computer-implemented method for generating s personalized neural network model includes accessing a shared or global neural network model. One or more personal inputs of a user are received. A set of features of the one or more inputs is extracted. An approximation of a posterior probability is computed based on the extracted features. A set of personalized weights are generated based on the approximated posterior probability. Processing one or more subsequent inputs via a personal model, including the set of personalized weights, enables generating of an inference.

公开(公告)号：US20250068662A1

公开(公告)日：2025-02-27

申请号：US18454456

申请日：2023-08-23

Applicant: QUALCOMM Incorporated

Inventor： Seokeon CHOI ,  Juntae LEE ,  Jaewon CHOI

IPC: G06F16/33

Abstract: Various embodiments include systems and methods for generating a privacy-aware multi-modal autoreply to an incoming communication. A processing system of a computing device may collect multi-modal information, determine a current user circumstance based on the collected information, determine a user privacy preference for autoreply responses, and generate a prompt that is input to a generative large language model (LLM) to generate optional autoreply responses, receive a list of personalized response suggestions from the generative LLM, and perform an autoreply action based on a selected personalized response suggestion.

公开(公告)号：US20240004889A1

公开(公告)日：2024-01-04

申请号：US18153899

申请日：2023-01-12

Applicant: QUALCOMM Incorporated

Inventor： Byeonggeun KIM ,  Juntae LEE ,  Simyung CHANG

IPC: G06F16/2458 ,  G06F16/28

CPC classification number: G06F16/2462 ,  G06F16/285

Abstract: Systems and techniques are provided for processing one or more data samples. For example, a neural network classifier can be trained to perform few-shot open-set recognition (FSOSR) based on a task-agnostic open-set prototype. A process can include determining one or more prototype representations for each class included in a plurality of support samples. A task-agnostic open-set prototype representation can be determined, in a same learned metric space as the one or more prototype representations. One or more distance metrics can be determined for each query sample of one or more query samples, based on the one or more prototype representations and the task-agnostic open-set prototype representation. Based on the one or more distance metrics, each query sample can be classified into one of classes associated with the one or more prototype representations or an open-set class associated with the task-agnostic open-set prototype representation.

公开(公告)号：US20250131262A1

公开(公告)日：2025-04-24

申请号：US18493286

申请日：2023-10-24

Applicant: QUALCOMM Incorporated

Inventor： Jihwan BANG ,  Kyu Woong HWANG ,  Simyung CHANG ,  Juntae LEE ,  Kyuhong SHIM ,  Seunghan YANG ,  Eunji KIM

IPC: G06N3/08

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for improved machine learning. A machine learning model is accessed. And an enrollment dataset for a device is accessed. A personalized model adapter generated based on the enrollment dataset and a plurality of model adapters is accessed. An input to the machine learning model is processed using the machine learning model in conjunction with the personalized model adapter. An output is provided, by the device, based on the processing.

公开(公告)号：US20240185088A1

公开(公告)日：2024-06-06

申请号：US18323197

申请日：2023-05-24

Applicant: QUALCOMM Incorporated

Inventor： Byeonggeun KIM ,  Juntae LEE ,  Seunghan YANG ,  Simyung CHANG

IPC: G06N3/0985 ,  G06N3/045 ,  G06N3/048

CPC classification number: G06N3/0985 ,  G06N3/045 ,  G06N3/048

Abstract: Certain aspects of the present disclosure provide techniques and apparatus for scalable weight reparameterization for efficient transfer learning. One example method generally includes training a first neural network to perform a task based on weights defined for a machine learning (ML) model trained to perform a different task and learned reparameterizing weights for each of a plurality of layers in the ML model; training a second neural network to generate a plurality of gating parameters based on a cost factor and the trained first neural network, each respective gating parameter of the plurality of gating parameters corresponding to weights in a respective layer of the plurality of layers; and updating the ML model based on the weights defined for the ML model, each gating parameter for each layer of the plurality of layers, and the learned reparameterizing weights for each layer of the plurality of layers.

公开(公告)号：US20240282081A1

公开(公告)日：2024-08-22

申请号：US18504968

申请日：2023-11-08

Applicant: QUALCOMM Incorporated

Inventor： Juntae LEE ,  Sungrack YUN

IPC: G06V10/764 ,  G06V10/44 ,  G06V10/62 ,  G06V10/82

CPC classification number: G06V10/764 ,  G06V10/44 ,  G06V10/62 ,  G06V10/82

Abstract: Systems and techniques are described herein for performing dynamic temporal fusion for video classification, such as recognition, detection, and/or other form of classification. For example, a computing device can generate, via a first network, frame-level features obtained from a set of input frames. The computing device can generate, via a first multi-scale temporal feature fusion engine, first local temporal context features from a first neighboring sub-sequence of the set of input frames. The computing device can generate, via a second multi-scale temporal feature fusion engine, second local temporal context features from a second neighboring sub-sequence of the set of input frames. The computing device can further classify the set of input frames based on the first local temporal context features and the second local temporal context features.

公开(公告)号：US20220318633A1

公开(公告)日：2022-10-06

申请号：US17705248

申请日：2022-03-25

Applicant: QUALCOMM Incorporated

Inventor： Jangho KIM ,  Simyung CHANG ,  Hyunsin PARK ,  Juntae LEE ,  Jaewon CHOI ,  Kyu Woong HWANG

IPC: G06N3/08 ,  G06N3/04

Abstract: A processor-implemented method for compressing a deep neural network model includes receiving an initial neural network model. The initial neural network is pruned based on a first threshold to generate a pruned network and a set of pruned weights. A quantization process is applied to the pruned network to produce a pruned and quantized network. A teacher model is generated by incorporating the pruned set of weights with the pruned network. In addition, an initial student model is generated from the quantized and pruned network. The initial student model is trained using the teacher model to output a trained student model.

公开(公告)号：US20220121949A1

公开(公告)日：2022-04-21

申请号：US17506646

申请日：2021-10-20

Applicant: QUALCOMM Incorporated

Inventor： Simyung CHANG ,  Jangho KIM ,  Hyunsin PARK ,  Juntae LEE ,  Jaewon CHOI ,  Kyu Woong HWANG

IPC: G06N3/08 ,  G06K9/62

Abstract: A method for generating a personalized model includes receiving one or more personal data samples from a user. A prototype of a personal identity is generated based on the personal data samples. The prototype of the personal identity is trained to reflect personal characteristics of the user. A network graph is generated based on the prototype of the personal identity. One or more channels of a global network are pruned based on the network graph to produce the personalized model.

公开(公告)号：US20240303987A1

公开(公告)日：2024-09-12

申请号：US18360741

申请日：2023-07-27

Applicant: QUALCOMM Incorporated

Inventor： Juntae LEE ,  Mihir JAIN ,  Sungrack YUN

IPC: G06V20/40 ,  G06F16/732 ,  G06F16/735 ,  G06F16/75

CPC classification number: G06V20/48 ,  G06F16/7328 ,  G06F16/735 ,  G06F16/75 ,  G06V20/41 ,  G06V10/82

Abstract: Aspects of the disclosure are directed to an apparatus configured to perform common-action localization. In certain aspects, the apparatus may receive a query video comprising a plurality of frames, wherein a first query proposal is determined based on a subset of frames of the plurality of frames, the first query proposal indicative of an action depicted on the subset of frames. In certain aspects, the apparatus may determine a first attendance for a first support video of a plurality of support videos. In certain aspects, the apparatus may determine a second attendance for a second support video of the plurality of support videos after computing the first attendance.