公开(公告)号：US20230409869A1

公开(公告)日：2023-12-21

申请号：US18316152

申请日：2023-05-11

Applicant: STMicroelectronics ( Rousset ) SAS

Inventor： Laurent Folliot ,  Pierre Demaj

IPC: G06N3/04

CPC classification number: G06N3/04

Abstract: According to one aspect, there is proposed a method for transforming a trained artificial neural network including a binary convolution layer followed by a pooling layer then a batch normalization layer, the method includes obtaining the trained artificial neural network and transforming the trained artificial neural network such that the order of the layers of the trained artificial neural network is modified by displacing the batch normalization layer after the convolution layer.

公开(公告)号：US11500767B2

公开(公告)日：2022-11-15

申请号：US16810546

申请日：2020-03-05

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Laurent Folliot ,  Pierre Demaj

IPC: G06F21/00 ,  G06F12/02

Abstract: In accordance with an embodiment, a method for determining an overall memory size of a global memory area configured to store input data and output data of each layer of a neural network includes: for each current layer of the neural network after a first layer, determining a pair of elementary memory areas based on each preceding elementary memory area associated with a preceding layer, wherein: the two elementary memory areas of the pair of elementary memory areas respectively have two elementary memory sizes, each of the two elementary memory areas are configured to store input data and output data of the current layer of the neural network, the output data is respectively stored in two different locations, and the overall memory size of the global memory area corresponds to a smallest elementary memory size at an output of the last layer of the neural network.

公开(公告)号：US20200302278A1

公开(公告)日：2020-09-24

申请号：US16810546

申请日：2020-03-05

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Laurent Folliot ,  Pierre Demaj

IPC: G06N3/063 ,  G06F12/02

Abstract: In accordance with an embodiment, a method for determining an overall memory size of a global memory area configured to store input data and output data of each layer of a neural network includes: for each current layer of the neural network after a first layer, determining a pair of elementary memory areas based on each preceding elementary memory area associated with a preceding layer, wherein: the two elementary memory areas of the pair of elementary memory areas respectively have two elementary memory sizes, each of the two elementary memory areas are configured to store input data and output data of the current layer of the neural network, the output data is respectively stored in two different locations, and the overall memory size of the global memory area corresponds to a smallest elementary memory size at an output of the last layer of the neural network.

公开(公告)号：US20200169631A1

公开(公告)日：2020-05-28

申请号：US16774954

申请日：2020-01-28

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Pierre Demaj ,  Laurent Folliot

IPC: H04M1/725 ,  G01D21/02 ,  H04W4/02 ,  H04W12/00 ,  H04W4/029 ,  H04W52/02 ,  G06F9/445 ,  G06F3/01 ,  G06K9/00

Abstract: A method of real-time scene detection performed by a wireless communication device includes, performing a first scene detection measurement to determine that the wireless communication device is located in a first scene. The first scene detection measurement is performed at first instant in time. The first scene is a type of environment. The method further includes associating the first scene with a corresponding reference scene of a predetermined set of reference scenes, determining a reference duration associated with the corresponding reference scene, and performing a second scene detection measurement immediately following expiration of the reference duration measured from the first instant in time.

公开(公告)号：US20180048355A1

公开(公告)日：2018-02-15

申请号：US15797947

申请日：2017-10-30

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Yoann Bouvet ,  Pierre Demaj

IPC: H04B3/54 ,  H04L25/02 ,  H04L5/00 ,  H04L27/26

CPC classification number: H04B3/544 ,  H04B3/466 ,  H04B3/542 ,  H04B2203/5495 ,  H04L5/006 ,  H04L25/023 ,  H04L27/26 ,  H04L27/2649

Abstract: A method is for processing a channel analog signal coming from a transmission channel. The method may include converting the channel analog signal into a channel digital signal, and detecting a state of the transmission channel based on the channel digital signal to detect whether the transmission channel is, over an interval of time, one or more of linear and time invariant and linear and cyclostationary.

公开(公告)号：US09838080B2

公开(公告)日：2017-12-05

申请号：US15042429

申请日：2016-02-12

Applicant: STMICROELECTRONICS (ROUSSET) SAS

Inventor： Yoann Bouvet ,  Pierre Demaj

IPC: H04B3/54 ,  H04B3/466 ,  H04L5/00 ,  H04L25/02 ,  H04L27/26

CPC classification number: H04B3/544 ,  H04B3/466 ,  H04B3/542 ,  H04B2203/5495 ,  H04L5/006 ,  H04L25/023 ,  H04L27/2649

Abstract: A method is for processing a channel analog signal coming from a transmission channel. The method may include converting the channel analog signal into a channel digital signal, and detecting a state of the transmission channel based on the channel digital signal to detect whether the transmission channel is, over an interval of time, one or more of linear and time invariant and linear and cyclostationary.

公开(公告)号：US20170288918A1

公开(公告)日：2017-10-05

申请号：US15629257

申请日：2017-06-21

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Mark Wallis ,  Yoann Bouvet ,  Pierre Demaj

IPC: H04B3/54 ,  H04L27/26 ,  H04L5/00 ,  H04L25/03 ,  H04L25/06

CPC classification number: H04B3/54 ,  H04L5/0007 ,  H04L25/03993 ,  H04L25/067 ,  H04L27/2607 ,  H04L27/2656 ,  H04L27/2675 ,  H04L27/2691

Abstract: A method is for processing an analog signal coming from a transmission channel. The analog signal may include a useful signal modulated on a sub-set of carriers. The method may include analog-to-digital converting of the analog signal into a digital signal, and synchronization processing the digital signal. The synchronizing may include determining, in a time domain, a limited number of coefficients of a predictive filter from an autoregressive model of the digital signal, and filtering the digital signal in the time domain by a digital finite impulse response filter with coefficients based upon the limited number of coefficients to provide a filtered digital signal. The method may include detecting of an indication allowing a location in the frame structure to be identified, using the filtered digital signal and a reference signal.

公开(公告)号：US20170180005A1

公开(公告)日：2017-06-22

申请号：US15454904

申请日：2017-03-09

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Pierre Demaj ,  Yoann Bouvet

IPC: H04B3/54 ,  H04L27/26 ,  H04L25/02

CPC classification number: H04B3/542 ,  H04B3/54 ,  H04B2203/5425 ,  H04L25/0202 ,  H04L25/0228 ,  H04L27/2647 ,  H04L27/2662

Abstract: A method is for processing an analog channel signal from a transmission channel. The method may include converting of the analog channel signal to a digital channel signal, and performing a channel estimation digital processing of the digital channel signal. The channel estimation digital processing may include for at least one frame, generating transfer functions of the transmission channel, the transfer functions respectively associated with reference symbols of the frame, and averaging processing of the transfer functions to generate an average transfer function. The method may include decoding of symbols of the frame following the reference symbols using the average transfer function.

公开(公告)号：US20240095502A1

公开(公告)日：2024-03-21

申请号：US18470281

申请日：2023-09-19

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Pierre Demaj ,  Laurent Folliot

IPC: G06N3/0464

CPC classification number: G06N3/0464

Abstract: An artificial neural network includes a unit cell. The unit cell includes a first binary two-dimensional convolution layer configured to receive an input tensor and to generate a first tensor. A first batch normalization layer is configured to receive the first tensor and to generate a second tensor. A concatenation layer is configured to generate a third tensor by concatenating the input tensor and the second tensor. A second binary two-dimensional convolution layer is configured to receive the third tensor and to generate a fourth tensor. A second batch normalization layer is configured to generate an output tensor based on the fourth tensor.

公开(公告)号：US20220164664A1

公开(公告)日：2022-05-26

申请号：US17510273

申请日：2021-10-25

Applicant: STMicroelectronics (Rousset) SAS

Inventor： Pierre Demaj ,  Laurent Folliot

IPC: G06N3/08 ,  G06F7/499 ,  G06F7/02

Abstract: According to one aspect, the disclosure proposes a method for updating an artificial neural network including initial weights stored in a memory at least in an integer format, which method includes: a processing unit determining the error gradients at the output of the layers of the neural network, the processing unit retrieving the initial weights from memory, the processing unit updating the initial weights comprising, for each initial weight, a first calculation of a corrected weight, in the integer format of this initial weight, the processing unit replacing the value of the initial weights stored in the memory by the value of the corrected weights.