公开(公告)号：US11741712B2

公开(公告)日：2023-08-29

申请号：US17463757

申请日：2021-09-01

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Farley Lai ,  Hans Peter Graf ,  Alexandru Niculescu-Mizil ,  Renqiang Min ,  Honglu Zhou

IPC: G06V20/40 ,  G06T7/73 ,  G06T7/246 ,  G06F18/213 ,  G06N3/045

CPC classification number: G06V20/41 ,  G06F18/213 ,  G06N3/045 ,  G06T7/246 ,  G06T7/73 ,  G06V20/46 ,  G06T2207/10016 ,  G06T2207/20081 ,  G06T2207/20084 ,  G06V2201/07

Abstract: A method for using a multi-hop reasoning framework to perform multi-step compositional long-term reasoning is presented. The method includes extracting feature maps and frame-level representations from a video stream by using a convolutional neural network (CNN), performing object representation learning and detection, linking objects through time via tracking to generate object tracks and image feature tracks, feeding the object tracks and the image feature tracks to a multi-hop transformer that hops over frames in the video stream while concurrently attending to one or more of the objects in the video stream until the multi-hop transformer arrives at a correct answer, and employing video representation learning and recognition from the objects and image context to locate a target object within the video stream.

公开(公告)号：US11736867B2

公开(公告)日：2023-08-22

申请号：US17579516

申请日：2022-01-19

Applicant: NEC Laboratories America, Inc.

Inventor： Junqiang Hu ,  Ting Wang

IPC: H04R23/00 ,  G01H9/00 ,  H03G3/30 ,  H04R1/08 ,  H04R3/00

CPC classification number: H04R23/008 ,  G01H9/004 ,  H03G3/3005 ,  H04R1/08 ,  H04R3/00 ,  H03G2201/103 ,  H04R2430/01

Abstract: Aspects of the present disclosure describe DFOS/DAS systems, methods, and structures that employ active microphones to enhance DAS operational capabilities by using an active circuit to amplify acoustic signals including voice(s). The circuit includes a microphone to collect acoustic signal(s) resulting from voice signals in the environment, and a speaker or a vibration device driven by an amplifier. The circuit can be clipped onto the fiber, with direct contact through the speaker or vibration device. A microcontroller may advantageously be employed to control the circuit for reduced power consumption, by detecting activities locally and only enabling the speaker when needed. The microcontroller may also send other information such as battery status to the DFOS interrogator through vibration codes.

公开(公告)号：US20230253068A1

公开(公告)日：2023-08-10

申请号：US18151686

申请日：2023-01-09

Applicant: NEC Laboratories America, Inc.

Inventor： Renqiang Min ,  Hans Peter Graf ,  Ziqi Chen

IPC: G16B15/30 ,  G06N20/00 ,  G16B20/50 ,  G16B40/20

CPC classification number: G16B15/30 ,  G06N20/00 ,  G16B20/50 ,  G16B40/20

Abstract: A method for implementing deep reinforcement learning with T-cell receptor (TCR) mutation policies to generate binding TCRs recognizing target peptides for immunotherapy is presented. The method includes extracting peptides to identify a virus or tumor cells, collecting a library of TCRs from target patients, predicting, by a deep neural network, interaction scores between the extracted peptides and the TCRs from the target patients, developing a deep reinforcement learning (DRL) framework with TCR mutation policies to generate TCRs with maximum binding scores, defining reward functions based on a reconstruction-based score and a density estimation-based score, randomly sampling batches of TCRs and following a policy network to mutate the TCRs, outputting mutated TCRs, and ranking the outputted TCRs to utilize top-ranked TCR candidates to target the virus or the tumor cells for immunotherapy.

公开(公告)号：US20230252139A1

公开(公告)日：2023-08-10

申请号：US18157180

申请日：2023-01-20

Applicant: NEC Laboratories America, Inc.

Inventor： Yanchi Liu ,  Xuchao Zhang ,  Haifeng Chen ,  Wei Cheng ,  Shengming Zhang

IPC: G06F21/55

CPC classification number: G06F21/554

Abstract: A method for implementing a self-attentive encoder-decoder transformer framework for anomaly detection in event sequences is presented. The method includes feeding event content information into a content-awareness layer to generate event representations, inputting, into an encoder, event sequences of two hierarchies to capture long-term and short-term patterns and to generate feature maps, adding, in the decoder, a special sequence token at a beginning of an input sequence under detection, during a training stage, applying a one-class objective to bound the decoded special sequence token with a reconstruction loss for sequence forecasting using the generated feature maps from the encoder, and during a testing stage, labeling any event representation whose decoded special sequence token lies outside a hypersphere as an anomaly.

公开(公告)号：US20230152791A1

公开(公告)日：2023-05-18

申请号：US17984413

申请日：2022-11-10

Applicant: NEC Laboratories America, Inc.

Inventor： LuAn Tang ,  Yuncong Chen ,  Wei Cheng ,  Haifeng Chen ,  Zhengzhang Chen ,  Yuji Kobayashi

IPC: G05B23/02 ,  G06N20/00

CPC classification number: G05B23/0221 ,  G06N20/00 ,  G05B23/0235 ,  G05B23/0237

Abstract: Systems and methods for defect detection for vehicle operations, including collecting a multiple modality input data stream from a plurality of different types of vehicle sensors, extracting one or more features from the input data stream using a grid-based feature extractor, and retrieving spatial attributes of objects positioned in any of a plurality of cells of the grid-based feature extractor. One or more anomalies are detected based on residual scores generated by each of cross attention-based anomaly detection and time-series-based anomaly detection. One or more defects are identified based on a generated overall defect score determined by integrating the residual scores for the cross attention-based anomaly detection and the time-series based anomaly detection being above a predetermined defect score threshold. Operation of the vehicle is controlled based on the one or more defects identified.

公开(公告)号：US20230152150A1

公开(公告)日：2023-05-18

申请号：US17987007

申请日：2022-11-15

Applicant: NEC Laboratories America, Inc.

Inventor： Yuheng CHEN ,  Ming-Fang HUANG ,  Ting WANG ,  Jingnan ZHAO

IPC: G01H9/00 ,  G01P3/36 ,  G06N20/10 ,  G06N3/08

CPC classification number: G01H9/004 ,  G01P3/36 ,  G06N20/10 ,  G06N3/08

Abstract: A fiber optic sensing cable located along a side of a paved road and runs parallel to a driving direction is monitored by distributed fiber optic sensing (DFOS) using Rayleigh backscattering generated along the length of the optical sensor fiber cable under dynamic vehicle loads. The interaction of vehicles with roadway locations exhibiting distressed pavement generates unique patterns of localized signals that are identified/distinguished from signals resulting from vehicles operating on roadway exhibiting a smooth pavement surface. Machine learning methods are employed to estimate an overall road surface quality as well as localizing pavement damage. Power spectral density estimation, principal component analysis, support vector machine (SVM) combined with principal component analysis (PCA), local binary pattern (LBP), and convolutional neural network (CNN) are applied to develop the machine learning models.

公开(公告)号：US20230146473A1

公开(公告)日：2023-05-11

申请号：US17967812

申请日：2022-10-17

Applicant: NEC Laboratories America, Inc.

Inventor： Ezra IP ,  Yue-Kai HUANG

IPC: G01H9/00 ,  G01D5/353 ,  G01D5/26 ,  H04L27/227

CPC classification number: G01H9/004 ,  G01D5/268 ,  G01D5/35361 ,  H04L27/2278

Abstract: Aspects of the present disclosure directed to frequency drift compensation for coded-DAS systems that use chirped pulses as a probe signal. Our inventive approach estimates timing jitter by correlating the amplitude of the estimated Rayleigh impulse response of every frame with a reference frame, and then re-aligns each frame by the estimated timing jitter. As the amount of timing jitter varies within a frame, every frame is divided into blocks where all samples have similar timing jitter, and perform timing jitter estimation and compensation on a block-by-block, frame-by-frame basis using an overlap-and-save method. Tracking of a slowly changing channel is enabled by allowing the reference frame to be periodically updated.

公开(公告)号：US20230125456A1

公开(公告)日：2023-04-27

申请号：US17968265

申请日：2022-10-18

Applicant: NEC Laboratories America, Inc.

Inventor： Yangmin DING ,  Yue TIAN ,  Sarper OZHARAR ,  Ting WANG

IPC: G08G1/04 ,  G01V1/22 ,  G01V1/00 ,  G01V1/30 ,  G08G1/01

Abstract: Aspects of the present disclosure describe dynamic road traffic noise mapping using DFOS over a telecommunications network that enables mapping of road traffic-induced noise at any observer location. DFOS is used to obtain instant traffic data including vehicle speed, volume, and vehicle types, based on vibration and acoustic signal along the length of a sensing fiber along with location information. A sound pressure level at a point of interest is determined, and traffic data associated with such point is incorporated into a reference noise emission database and a wave propagation theory for total sound pressure level prediction and mapping. Real-time wind speed using DFOS—such as distributed acoustic sensing (DAS)—is obtained to provide sound pressure adjustment due to the wind speed.

公开(公告)号：US20230112441A1

公开(公告)日：2023-04-13

申请号：US17961169

申请日：2022-10-06

Applicant: NEC Laboratories America, Inc. ,  NEC Corporation

Inventor： LuAn Tang ,  Yuncong Chen ,  Wei Cheng ,  Zhengzhang Chen ,  Haifeng Chen ,  Yuji Kobayashi ,  Yuxiang Ren

IPC: B60W40/09 ,  G06K9/62

Abstract: Systems and methods for data fusion and analysis of vehicle sensor data, including receiving a multiple modality input data stream from a plurality of different types of vehicle sensors, determining latent features by extracting modality-specific features from the input data stream, and aligning a distribution of the latent features of different modalities by feature-level data fusion. Classification probabilities can be determined for the latent features using a fused modality scene classifier. A tree-organized neural network can be trained to determine path probabilities and issue driving pattern judgments, with the tree-organized neural network including a soft tree model and a hard decision leaf. One or more driving pattern judgments can be issued based on a probability of possible driving patterns derived from the modality-specific features.

公开(公告)号：US20230086023A1

公开(公告)日：2023-03-23

申请号：US17940599

申请日：2022-09-08

Applicant: NEC Laboratories America, Inc.

Inventor： Farley Lai ,  Asim Kadav ,  Cheng-En Wu

IPC: G06V20/40 ,  G06V10/771

Abstract: A method for model training and deployment includes training, by a processor, a model to learn video representations with a self-supervised contrastive loss by performing progressive training in phases with an incremental number of positive instances from one or more video sequences, resetting the learning rate schedule in each of the phases, and inheriting model weights from a checkpoint from a previous training phase. The method further includes updating the trained model with the self-supervised contrastive loss given multiple positive instances obtained from Cascade K-Nearest Neighbor mining of the one or more video sequences by extracting features in different modalities to compute similarities between the one or more video sequences and selecting a top-k similar instances with features in different modalities. The method also includes fine-tuning the trained model for a downstream task. The method additionally includes deploying the trained model for a target application inference for the downstream task.