公开(公告)号：US20250061334A1

公开(公告)日：2025-02-20

申请号：US18805978

申请日：2024-08-15

Applicant: NEC Laboratories America, Inc.

Inventor： Yanchi Liu ,  Wei Cheng ,  Xujiang Zhao ,  Runxue Bao ,  Haifeng Chen ,  Nan Zhang

IPC: G06N3/082 ,  G06N3/0455

Abstract: Systems and methods for optimizing large language models (LLM) with domain-oriented model compression. Importance weights for general knowledge in a trained LLM, pretrained with deep learning, can be determined by computing the error when removing a weight from the trained LLM. The trained LLM can be iteratively optimized to obtain a domain-compressed LLM with domain knowledge while maintaining general knowledge by: fine-tuning the trained LLM iteratively with domain knowledge using the importance weights for general knowledge to obtain a fine-tuned LLM; determining importance weights for domain knowledge in the LLM with a regularization term by using gradient descent to optimize parameters when the fine-tuned LLM is trained with domain knowledge; and pruning learned knowledge based on importance weights for domain knowledge. A corrective action can be performed on a monitored entity using the domain-compressed LLM.

公开(公告)号：US12206489B2

公开(公告)日：2025-01-21

申请号：US18109247

申请日：2023-02-13

Applicant: NEC Laboratories America, Inc.

Inventor： Yue-Kai Huang ,  Ezra Ip

IPC: H04J14/02

Abstract: Aspects of the present disclosure describe distributed fiber optic sensing (DFOS) systems, methods, and structures that advantageously provide DFOS and WDM communications over amplified, multi-span optical WDM optical telecommunications facilities using all Raman amplification and coding schemes. Our all-Raman amplification operates stably—without isolators—and provides sufficient gain to compensate for fiber span loss for both DFOS signals and WDM channel signals—at the same time. Furthermore, our inventive techniques employ signal coding, such as MB-TGD-OFDR for DAS, and we operate our DFOS operation power at a much lower power level as compared to pulse interrogation techniques. With improved OSNR and reduced power using signal coding along with our distributed Raman amplification, our DFOS systems can co-exist with WDM communication channels on the same amplified multi-span fiber optic links over great distances.

公开(公告)号：US12205357B2

公开(公告)日：2025-01-21

申请号：US17715901

申请日：2022-04-07

Applicant: NEC Laboratories America, Inc.

Inventor： Shaobo Han ,  Renqiang Min ,  Tingfeng Li

IPC: G06V10/778 ,  G06V10/82 ,  G06V30/19

Abstract: A reinforcement learning based approach to the problem of query object localization, where an agent is trained to localize objects of interest specified by a small exemplary set. We learn a transferable reward signal formulated using the exemplary set by ordinal metric learning. It enables test-time policy adaptation to new environments where the reward signals are not readily available, and thus outperforms fine-tuning approaches that are limited to annotated images. In addition, the transferable reward allows repurposing of the trained agent for new tasks, such as annotation refinement, or selective localization from multiple common objects across a set of images. Experiments on corrupted MNIST dataset and CU-Birds dataset demonstrate the effectiveness of our approach.

公开(公告)号：US12205028B2

公开(公告)日：2025-01-21

申请号：US17958597

申请日：2022-10-03

Applicant: NEC Laboratories America, Inc.

Inventor： Yuncong Chen ,  Zhengzhang Chen ,  Xuchao Zhang ,  Wenchao Yu ,  Haifeng Chen ,  LuAn Tang ,  Zexue He

IPC: G06F40/30 ,  G06F40/47 ,  G06N3/08

Abstract: A computer-implemented method for multi-model representation learning is provided. The method includes encoding, by a trained time series (TS) encoder, an input TS segment into a TS-shared latent representation and a TS-private latent representation. The method further includes generating, by a trained text generator, a natural language text that explains the input TS segment, responsive to the TS-shared latent representation, the TS-private latent representation, and a text-private latent representation.

公开(公告)号：US12205027B2

公开(公告)日：2025-01-21

申请号：US17840987

申请日：2022-06-15

Applicant: NEC Laboratories America, Inc.

Inventor： Christopher Malon

IPC: G06F17/00 ,  G06F40/166 ,  G06F40/258 ,  G06F40/40 ,  G06N3/08

Abstract: A method for neural network training is provided. The method inputs a training set of textual claims, lists of evidence including gold evidence chains, and claim labels labelling the evidence with respect to the textual claims. The claim labels include refutes, supports, and not enough information (NEI). The method computes an initial set of document retrievals for each of the textual claims. The method also includes computing an initial set of page element retrievals including sentence retrievals from the initial set of document retrievals for each of the textual claims. The method creates, from the training set of textual claims, a Leave Out Training Set which includes input texts and target texts relating to the labels. The method trains a sequence-to-sequence neural network to generate new target texts from new input texts using the Leave Out Training Set.

公开(公告)号：US20250008132A1

公开(公告)日：2025-01-02

申请号：US18755150

申请日：2024-06-26

Applicant: NEC Laboratories America, Inc.

Inventor： Biplob Debnath ,  Deep Patel ,  Srimat Chakradhar ,  Christoph Reich

IPC: H04N19/33 ,  H04N19/124 ,  H04N19/176 ,  H04N19/186 ,  H04N19/625

Abstract: Systems and methods are provided for encoding and decoding images using differentiable JPEG compression, including converting images from RGB color space to YCbCr color space to obtain a luminance and chrominance channels, and applying chroma subsampling to the chrominance channels to reduce resolution. The YCbCr image is divided into pixel blocks and a DCT is performed on the pixel blocks to obtain DCT coefficients. DCT coefficients are quantized using a scaled quantization table to reduce precision, and quantized DCT coefficients are encoded using lossless entropy coding, forming a compressed JPEG file decoded by reversing the lossless entropy coding to obtain quantized DCT coefficients, which are dequantized using the scaled quantization table to restore the precision. The dequantized DCT coefficients are converted back to a spatial domain using an IDCT, the chrominance channels are upsampled to original resolution, and the YCbCr image is converted back to the RGB color space.

公开(公告)号：US12160090B2

公开(公告)日：2024-12-03

申请号：US18492902

申请日：2023-10-24

Applicant: NEC Laboratories America, Inc.

Inventor： Yangmin Ding ,  Yuanda Xu ,  Sarper Ozharar ,  Yue Tian ,  Ting Wang

IPC: H02G1/02 ,  G01D5/353 ,  G01S19/01

Abstract: Systems and methods for performing the dynamic anomaly localization of utility pole aerial/suspended/supported wires/cables by distributed fiber optic sensing. In sharp contrast to the prior art, our inventive systems and methods according to aspects of the present disclosure advantageously identify a “location region” on a utility pole supporting an affected wire/cable, thereby permitting the identification and reporting of service personnel that are uniquely responsible for responding to such anomalous condition(s).

公开(公告)号：US20240394524A1

公开(公告)日：2024-11-28

申请号：US18670275

申请日：2024-05-21

Applicant: NEC Laboratories America, Inc.

Inventor： Iain Melvin

IPC: G06N3/063 ,  G16H20/00 ,  G16H50/20

Abstract: Methods and systems for configuring a machine learning model include selecting a head from a set of stored heads, responsive to an input, to implement a layer in a transformer machine learning model. The selected head is copied from persistent storage to active memory. The layer in the transformer machine learning model is executed on the input using the selected head to generate an output. An action is performed responsive to the output.

公开(公告)号：US20240371521A1

公开(公告)日：2024-11-07

申请号：US18649072

申请日：2024-04-29

Applicant: NEC Laboratories America, Inc.

Inventor： Wenchao Yu ,  Haifeng Chen ,  Wei Cheng

IPC: G16H50/20 ,  G06N3/098

Abstract: Methods and systems for skill prediction include aggregating locally trained parameters from client systems to generate updated global parameters. Parameterized vectors from the client systems are clustered into prototype clusters. A centroid of each prototype cluster is determined and the parameterized vectors from the client systems are matched to centroids of the prototype clusters to identify sets of updated local prototype vectors. The updated global parameters and the updated local prototype vectors are distributed to the client systems.

公开(公告)号：US12136255B2

公开(公告)日：2024-11-05

申请号：US17577664

申请日：2022-01-18

Applicant: NEC Laboratories America, Inc.

Inventor： Yi Yang ,  Murugan Sankaradas ,  Srimat Chakradhar

IPC: G06K9/00 ,  G06V10/774 ,  H04L67/00 ,  H04L67/01

Abstract: A method for employing a semi-supervised learning approach to improve accuracy of a small model on an edge device is presented. The method includes collecting a plurality of frames from a plurality of video streams generated from a plurality of cameras, each camera associated with a respective small model, each small model deployed in the edge device, sampling the plurality of frames to define sampled frames, performing inference to the sampled frames by using a big model, the big model shared by all of the plurality of cameras and deployed in a cloud or cloud edge, using the big model to generate labels for each of the sampled frames to generate training data, and training each of the small models with the training data to generate updated small models on the edge device.