公开(公告)号：US20200334539A1

公开(公告)日：2020-10-22

申请号：US16728987

申请日：2019-12-27

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Yu Wang ,  Yilin Shen ,  Yue Deng ,  Hongxia Jin

IPC: G06N3/08 ,  G06N3/04 ,  G06N20/20 ,  G06K9/62

Abstract: Intent determination based on one or more multi-model structures can include generating an output from each of a plurality of domain-specific models in response to a received input. The domain-specific models can comprise simultaneously trained machine learning models that are trained using a corresponding local loss metric for each domain-specific model and a global loss metric for the plurality of domain-specific models. The presence or absence of an intent corresponding to one or more domain-specific models can be determined by classifying the output of each domain-specific model.

公开(公告)号：US09686397B1

公开(公告)日：2017-06-20

申请号：US15280325

申请日：2016-09-29

Applicant: Samsung Electronics Co., Ltd.

Inventor： Pengfei Hu ,  Yilin Shen ,  Hongxia Jin

IPC: H04B11/00 ,  H04B1/38 ,  H04M1/725 ,  H04L12/26 ,  H04W72/04

CPC classification number: H04B11/00 ,  Y02D70/142 ,  Y02D70/144 ,  Y02D70/26 ,  Y02D70/48

Abstract: A modulation method referred to as Time Shift Keying (TSK) is used to transmit messages between two devices in a highly energy efficient manner. A message represented by an inaudible audio signal is modulated on a transmitting device. The audio signal is comprised of an array of non-zero amplitude delimiter signals with time periods of zero-amplitude transmission between delimiters. The time duration of the zero-amplitude transmission periods is mapped to a symbol, multiple symbols are then assembled into a message. On the transmitting device, the audio signal is broken into pieces or sequences of bits which are mapped to symbols. On the receiving device, the time durations of zero-amplitude transmission are translated to the symbols which are assembled to the message. The delimiter signals have gradually increasing and decreasing amplitudes and have a length such that make them detectable by the receiving device.

公开(公告)号：US20170109544A1

公开(公告)日：2017-04-20

申请号：US15294570

申请日：2016-10-14

Applicant: Samsung Electronics Co., Ltd

Inventor： Rui Chen ,  Yilin Shen ,  Hongxia Jin

IPC: G06F21/62 ,  H04L29/06

CPC classification number: G06F21/6254 ,  G06F21/6245 ,  H04L63/0407 ,  H04L63/0421

Abstract: An apparatus, method, and computer readable medium for management of infinite data streams. The apparatus includes a memory that stores streaming data with a data set and a processor operably connected to the memory. The processor transforms the data set to a second data set. To transform the data set, the processor determines whether a difference level exceeds a threshold, and transforms the data set by adding a noise when the difference level exceeds the threshold. When the difference level does not exceed the threshold, the processor determines whether a retroactive count is greater than a threshold, transforms the data set by adding a second noise when the retroactive count is greater than the threshold, and transforms the data set by adding a third noise when the retroactive count is not greater than the threshold. The processor transmits the second data set to a data processing system for further processing.

公开(公告)号：US20250021826A1

公开(公告)日：2025-01-16

申请号：US18629162

申请日：2024-04-08

Applicant: Samsung Electronics Co., Ltd.

Inventor： Shangqian Gao ,  Ting Hua ,  Yen-Chang Hsu ,  Yilin Shen ,  Hongxia Jin

IPC: G06N3/0985 ,  G06N3/0495

Abstract: In one embodiment, a method includes accessing at least a portion of a training dataset for a trained neural network that includes multiple layers, where each layer includes a number of parameters, and where the training dataset includes multiple training samples that each include an input and a ground-truth output used to train the trained neural network. The method further includes training a hypernetwork to generate a layer-specific compression mask for each of one or more of the multiple layers of the trained neural network. The method further includes generating, by the trained hypernetwork, a final layer-specific compression mask for the trained neural network and compressing the trained neural network by reducing, for each of the one or more layers of the neural network, the number of parameters of that layer according to the final layer-specific compression mask.

公开(公告)号：US12183062B2

公开(公告)日：2024-12-31

申请号：US17589535

申请日：2022-01-31

Applicant: Samsung Electronics Co., Ltd.

Inventor： Changsheng Zhao ,  Burak Uzkent ,  Yilin Shen ,  Hongxia Jin

IPC: G06V10/80 ,  G06F40/279 ,  G06V10/774 ,  G06V10/778

Abstract: A method includes obtaining a batch of training data including multiple paired image-text pairs and multiple unpaired image-text pairs, where each paired image-text pair and each unpaired image-text pair includes an image and a text. The method also includes training a machine learning model using the training data based on an optimization of a combination of losses. The losses include, for each paired image-text pair, (i) a first multi-modal representation loss based on the paired image-text pair and (ii) a second multi-modal representation loss based on two or more unpaired image-text pairs, selected from among the multiple unpaired image-text pairs, wherein each of the two or more unpaired image-text pairs includes either the image or the text of the paired image-text pair.

公开(公告)号：US20240394592A1

公开(公告)日：2024-11-28

申请号：US18434691

申请日：2024-02-06

Applicant: Samsung Electronics Co., Ltd.

Inventor： Rakshith Sharma Srinivasa ,  Jaejin Cho ,  Chouchang Yang ,  Yashas Malur Saidutta ,  Ching-Hua Lee ,  Yilin Shen ,  Hongxia Jin

IPC: G06N20/00

Abstract: A method includes accessing a training dataset having multiple samples, where each sample includes a data point for each of multiple modalities. The method also includes generating, using a first encoder associated with a first modality of the multiple modalities, first modality embeddings for data points of the first modality in the training dataset. The method further includes, for each first modality embedding, determining a similarity metric to other first modality embeddings. The method also includes generating, using a second encoder associated with a second modality of the multiple modalities, second modality embeddings for data points of the second modality in the training dataset. In addition, the method includes training the second encoder based on a contrastive loss function to align the first modality embeddings and the second modality embeddings from different samples of the training dataset, where the contrastive loss function is weighed using the similarity metrics.

公开(公告)号：US20240046946A1

公开(公告)日：2024-02-08

申请号：US18058104

申请日：2022-11-22

Applicant: Samsung Electronics Co., Ltd.

Inventor： Chou-Chang Yang ,  Ching-Hua Lee ,  Rakshith Sharma Srinivasa ,  Yashas Malur Saidutta ,  Yilin Shen ,  Hongxia Jin

IPC: G10L21/0232 ,  G10L15/06 ,  G10L15/02 ,  G10L25/18

CPC classification number: G10L21/0232 ,  G10L15/063 ,  G10L15/02 ,  G10L25/18 ,  G10L2021/02166

Abstract: A method includes obtaining, using at least one processing device, noisy speech signals and extracting, using the at least one processing device, acoustic features from the noisy speech signals. The method also includes receiving, using the at least one processing device, a predicted speech mask from a speech mask prediction model based on a first acoustic feature subset and receiving, using the at least one processing device, a predicted noise mask from a noise mask prediction model based on a second acoustic feature subset. The method further includes providing, using the at least one processing device, predicted speech features determined using the predicted speech mask and predicted noise features determined using the predicted noise mask to a filtering mask prediction model. In addition, the method includes generating, using the at least one processing device, a clean speech signal using a predicted filtering mask output by the filtering mask prediction model.

公开(公告)号：US11854528B2

公开(公告)日：2023-12-26

申请号：US17402045

申请日：2021-08-13

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Yen-Chang Hsu ,  Yilin Shen ,  Avik Ray ,  Hongxia Jin

IPC: G06N20/00 ,  G10L15/02 ,  G10L15/18

CPC classification number: G10L15/02 ,  G10L15/18

Abstract: An apparatus for detecting unsupported utterances in natural language understanding, includes a memory storing instructions, and at least one processor configured to execute the instructions to classify a feature that is extracted from an input utterance of a user, as one of in-domain and out-of-domain (OOD) for a response to the input utterance, obtain an OOD score of the extracted feature, and identify whether the feature is classified as OOD. The at least one processor is further configured to executed the instructions to, based on the feature being identified to be classified as in-domain, identify whether the obtained OOD score is greater than a predefined threshold, and based on the OOD score being identified to be greater than the predefined threshold, re-classify the feature as OOD.

公开(公告)号：US11720814B2

公开(公告)日：2023-08-08

申请号：US16234433

申请日：2018-12-27

Applicant: Samsung Electronics Co., Ltd.

Inventor： Yilin Shen ,  Yue Deng ,  Hongxia Jin

IPC: G06N20/00 ,  G06F18/24 ,  G06V10/764 ,  G06N3/049

CPC classification number: G06N20/00 ,  G06F18/24 ,  G06V10/764 ,  G06N3/049

Abstract: A recognition method includes retrieving an input including data of a first window size. The method further includes classifying the input based on comparison of warping distance of the input with a pruning threshold.

公开(公告)号：US20230177338A1

公开(公告)日：2023-06-08

申请号：US18073383

申请日：2022-12-01

Applicant: Samsung Electronics Co., Ltd.

Inventor： Qian Lou ,  Yen-Chang Hsu ,  Burak Uzkent ,  Ting Hua ,  Yilin Shen ,  Hongxia Jin

IPC: G06N3/082 ,  G06V10/82 ,  G06V10/772

CPC classification number: G06N3/082 ,  G06V10/82 ,  G06V10/772

Abstract: A method includes obtaining, using a first electronic device, a weight matrix associated with a trained transformer model. The method also includes factorizing the weight matrix into a dictionary weight matrix and an intermediate matrix. The method further includes pruning the intermediate matrix to generate a sparse intermediate matrix. The method also includes fine-tuning the sparse intermediate matrix based on a training dataset to generate a fine-tuned sparse intermediate matrix. The method further includes determining an index matrix and a coefficient matrix based on the fine-tuned sparse intermediate matrix. In addition, the method includes deploying the dictionary weight matrix, the index matrix, and the coefficient matrix to a second electronic device without deploying the weight matrix to the second electronic device. A number of parameters in the dictionary weight matrix, the index matrix, and the coefficient matrix is smaller than a number of parameters in the weight matrix.