EP4154189A1 - Procédés d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée, et d'apprentissage de paramètres d'un deuxième réseau de neurones - Google Patents

Procédés d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée, et d'apprentissage de paramètres d'un deuxième réseau de neurones

Info

Publication number
EP4154189A1
EP4154189A1 EP21732457.3A EP21732457A EP4154189A1 EP 4154189 A1 EP4154189 A1 EP 4154189A1 EP 21732457 A EP21732457 A EP 21732457A EP 4154189 A1 EP4154189 A1 EP 4154189A1
Authority
EP
European Patent Office
Prior art keywords
neural network
input
approximating
target layer
identity
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
EP21732457.3A
Other languages
German (de)
English (en)
French (fr)
Inventor
Hervé Chabanne
linda GUIGA
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
IDEMIA Public Security SAS
Original Assignee
Idemia Identity and Security France SAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Idemia Identity and Security France SAS filed Critical Idemia Identity and Security France SAS
Publication of EP4154189A1 publication Critical patent/EP4154189A1/fr
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/50—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems
    • G06F21/52—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems during program execution, e.g. stack integrity ; Preventing unwanted data erasure; Buffer overflow
    • G06F21/54—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems during program execution, e.g. stack integrity ; Preventing unwanted data erasure; Buffer overflow by adding security routines or objects to programs
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/09—Supervised learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn

Definitions

  • TITLE Methods of secure use of a first neural network on input data, and of learning parameters of a second neural network
  • the present invention relates to the field of artificial intelligence, and in particular to a method of secure use of a first neural network on input data.
  • Neural networks are widely used for the classification of data.
  • a neural network After a phase of machine learning (generally supervised, that is to say on an already classified reference database), a neural network "learns" and becomes on its own capable of applying the same classification to data. unknown. More precisely, the value of the weight and parameters of the NN is gradually modified until the latter is able to carry out the intended task.
  • a phase of machine learning generally supervised, that is to say on an already classified reference database
  • a first way is to increase the size, the number of layers and the number of parameters of the network so as to complicate the task of the attacker. If it works, on the one hand it only slows down the attacker and above all it degrades performance because the neural network is then unnecessarily heavy and difficult to learn.
  • a second avenue is to limit the number of entries that can be submitted to the neural network, or at least to detect suspicious sequences of entries. This is not always applicable, however, since the attacker can legally gain access to the neural network by having paid for unrestricted access, for example.
  • the present invention relates to a method of secure use of a first neural network on an input data item, the method being characterized in that it comprises the implementation by means of data processing of a terminal of steps of:
  • Said convolutional neural network is inserted at the input of a target layer of the first neural network.
  • Said target layer is a layer inside the first neural network.
  • Said convolutional neural network has an output size smaller than an input size of said target layer so as to approximate only certain input channels of this target layer.
  • Step (a) comprises selecting said target layer of the first neural network from among the layers of said first neural network.
  • Step (a) comprises selecting the input channels of said target layer to be approximated from among all the input channels of the target layer.
  • the at least one convolutional neural network approximating the identity function has an output size equal to the product of two integers.
  • the method comprises a preliminary step (aO) of obtaining the parameters of the first neural network and of the at least one convolutional neural network approximating the identity function.
  • Step (aO) includes obtaining the parameters of a convolutional neural network set approximating the identity function.
  • Step (a) comprises selecting from said set of at least one convolutional neural network approximating the identity function to be inserted.
  • Step (a) comprises, for each convolutional neural network approximating the identity function selected, said selection of said target layer of the first neural network from among the layers of said first neural network and / or selection of the input channels of said target layer to be approximated among all the input channels of the target layer.
  • Step (a) further comprises the prior selection of a number of convolutional neural networks approximating the identity function of said set to be selected.
  • Step (aO) is a step, implemented by data processing means of a server, of learning the parameters of the first neural network and of the at least one convolutional neural network approximating the function identity from at least one training database.
  • the first neural network and the convolutional neural network (s) approximating the identity function comprise an alternation of linear layers and activating non-linear layers.
  • Said activation function is the ReLU function.
  • Said target layer is a linear layer of the first neural network.
  • the at least one convolutional neural network approximating the identity function comprises two or three linear layers.
  • the linear layers of the convolutional neural network are filter convolution layers, for example of size 5 ⁇ 5.
  • a method of learning parameters of a second neural network is proposed, the method being characterized in that it comprises the implementation by means of data processing of a server of steps of : (a) construction of the second neural network corresponding to a first neural network into which is inserted at least one convolutional neural network approximating the identity function;
  • a method of secure use of a first neural network on input data comprising the learning of parameters of a second neural network in accordance with the precedent according to the second aspect; and the implementation by data processing means of a terminal of a step (b) of using the second neural network on said input data.
  • the invention relates to a computer program product comprising code instructions for the execution of a method according to the first or the third aspect of secure use of a first neural network on input data, or according to the second aspect of learning parameters of a second neural network; and a storage means readable by a computer equipment on which a computer program product comprises code instructions for the execution of a method according to the first or the third aspect of secure use of a first neural network on an input data item, or according to the second aspect of learning parameters of a second neural network.
  • FIG. 1 is a diagram of an architecture for the implementation of the methods according to the invention.
  • FIG. 2a schematically represents the steps of a first embodiment of a method of secure use of a first neural network on input data according to the invention
  • FIG. 2b schematically represents the steps of a second embodiment of a method of secure use of a first neural network on input data according to the invention
  • FIG. 3 schematically represents an example of the architecture of a second neural network encountered in the implementation of the methods according to the invention.
  • the server 1 is the learning equipment (implementing the second method) and the terminal 2 is a user equipment (implementing the first method. ).
  • Said method of use is implemented on an input datum, and is for example a classification of the input datum among several classes if it is a classification NN (but this task is not necessarily a classification even if it is the most classic).
  • NN classification NN
  • ReLU Rectified Linear Unit
  • each item of equipment 1, 2 is typically remote computer equipment connected to a wide area network 10 such as the Internet network for the exchange of data.
  • a wide area network 10 such as the Internet network for the exchange of data.
  • Each comprises data processing means 11, 21 of processor type, and data storage means 12, 22 such as a computer memory, for example a hard disk.
  • Server 1 stores a learning database, i.e. a set of data for which we already know the associated output, for example already classified (as opposed to the so-called input data that we are trying to process). It can be a high business value learning base that one seeks to keep secret.
  • a learning database i.e. a set of data for which we already know the associated output, for example already classified (as opposed to the so-called input data that we are trying to process). It can be a high business value learning base that one seeks to keep secret.
  • equipment 1 and 2 could be the same equipment, or even the learning base could be a public base.
  • the present method is not limited to a type of NN and therefore not to a particular nature of data, the input or training data may be representative of images, sounds, etc.
  • the 1 st NN can quite be a CNN, even if we will describe later a specialized CNN that we will use in the context of the present process.
  • these are biometric data, the input or training data being typically representative of images or even directly images of biometric features (faces, fingerprints, irises, etc.), or directly. preprocessed data from the biometric features (for example the position of minutiae in the case of fingerprints).
  • the present invention proposes to make the task of attackers more complex without complicating the NN by virtue of artificial hyperplanes.
  • we secure the NN by making it significantly more robust without making it heavier and degrading its performance.
  • first neural network will denote the original NN to be protected and “second neural network” the modified and thus secured NN.
  • securing the 1 st NN can be done a posteriori (once it has been learned), or from the start (ie we directly learn a secure version of the NN) .
  • securing the first network as a second network consists in integrating into its architecture at least one convolutional neural network (CNN) approximating the identity function (the latter will be referred to as “CNN Identity” for convenience).
  • CNN convolutional neural network
  • This “parasitic” CNN does not modify the operation of the NN because its outputs are substantially equal to its inputs. On the other hand, it breaks the original hyperplane structure.
  • FIG. 2a a first embodiment of the method of secure use of the 1 st NN on input data, implemented by the data processing means 21 of the terminal 2.
  • the method advantageously begins with a “preparatory” step (aO) of obtaining the parameters of the 1 st NN and of at least one CNN Identity, preferably a plurality of CNN Identities, in particular of various architectures, of various input sizes and outing, learned on different bases, etc., so as to define an if possible varied set of CNN Identity, we will see this in more detail later.
  • aO a “preparatory” step of obtaining the parameters of the 1 st NN and of at least one CNN Identity, preferably a plurality of CNN Identities, in particular of various architectures, of various input sizes and outing, learned on different bases, etc., so as to define an if possible varied set of CNN Identity, we will see this in more detail later.
  • This step (aO) can be a step of learning each of the networks on a dedicated learning base, in particular of the 1 st NN, preferably implemented by the data processing means 11 of the server 1 for this purpose, but it will be understood that the networks (in particular the CNN Identity) could be pre-existing and taken as is.
  • the CNN Identity (s) can be learned in particular on any public image database, or even on random data (no need for them to be annotated because it is assumed that the entry is also the expected output, possibly up to a noise, see below).
  • An alternative embodiment will be seen later in which there is no this step (aO).
  • step (a) said 2 nd NN corresponding to the 1 st NN in which is inserted at least one convolutional neural network approximating the identity function, in particular one or more selected Identity CNNs, is constructed.
  • step (a) is a step of inserting the CNN Identity (s) into the 1 st NN. If there are more than one CNN Identity selected, they can be inserted one after the other.
  • step (a) advantageously comprises the selection of one or more CNN Identity from said set of CNN Identity, for example randomly.
  • Other “insertion parameters” can be selected, in particular a position in the 1 st NN (target layer) and / or channels of a target layer of the 1 st NN, see below.
  • the set of CNN Identity contains only one CNN so that there is no need for selection, or even that the CNN Identity is learned on the fly.
  • the term “insertion” is understood to mean the addition of the layers of the CNN Identity upstream of the “target” layer of the 1 st NN so that the input of this layer is at least in part the output of the CNN Identity.
  • the CNN Identity "intercepts" all or part of the input of the target layer to replace it with its output. It is understood that since the CNN Identity approximates the identity function, its outputs are substantially identical to its inputs so that the data that the target layer receives are substantially identical to those intercepted.
  • the target layer is preferably a linear layer (and not a non-linear layer with an activation function for example), so the CNN Identity is inserted at the input of a linear layer of the 1 st NN.
  • the target layer is preferably a layer inside the 1 st NN, that is to say a layer other than the first (between the second layer and the last). Particularly preferably, the target layer is thus a linear layer inside the 1 st NN.
  • the CNN Identity has an output size smaller than an input size of said linear layer so as to approximate only certain input channels of this linear layer (i.e. not all).
  • input / output size is meant the number of input / output channels.
  • FIG. 3 represents a 1 st NN with three linear layers (including a central hidden layer), in which a CNN Identity is placed at the input of the second layer.
  • the first layer has eight input channels (size 8), while the identity CNN has only four input / output channels (by definition a CNN approximating the identity function at the same input and output dimensions ).
  • the eight input channels of the first linear layer only four are approximated, and the other four are as is. Affecting only some of the channels (i.e.
  • Step (a) can also comprise, as explained, the selection of the target layer and / or of the input channels of the target layer affected by the CNN Identity (if necessary previously selected). For example, in figure 3 these are channels 1, 2, 3 and 4, but we could have taken any set of four of the eight channels, for example channels 1, 3, 5 and 7. This selection can again be made randomly and dynamically, i.e.
  • new channels are drawn for each new request for use of the 1 st NN, but also according to a sequence, or else according to contextual data, and in particular to the data d 'Entrance.
  • the present invention will not be limited to any way of selecting in particular the target layer / the channels to approximate / the CNN Identity (s), as long as there is an active choice among several possibilities so as to add entropy to make the task ever more complex for a possible attacker .
  • the selection can be made according to the following protocol (each step being optional - each choice can be random or predetermined):
  • a target layer to be assigned ie at the input of which the CNN will be inserted is chosen from among the layers (in particular linear and / or inside) of the 1 st NN;
  • two CNN Identities can be chosen as affecting the same target layer: either the channels concerned are distinct and there is no problem, or at least one channel overlaps and in this case one can either decide that it is not desirable (and one start again the pulling), or accept that one CNN identity is upstream of the other: a channel can thus be approximated twice in a row before to arrive at the input of the target layer.
  • CNN Identity are typically networks working on images, ie two-dimensional objects (“rectangles”), and therefore presenting an equal number of input / output channels. to the product of two integers, ie of the form a * b, where a and b are integers each greater than or equal to two, and even preferably “squares” of dimension a 2 . It is quite possible to imagine using CNNs working on three-dimensional objects and therefore presenting a number of input / output channels equal to the product of three integers, ie of the form a * b * c, etc. In the example of figure 3, we have a CNN Identity working on 2x2 images, and therefore with four input / output channels.
  • step (a) it is assumed that the 2 nd NN has been constructed.
  • this 2 nd NN can be used on said input datum, ie we apply the 2 nN NN to the input datum and we obtain an output datum which can be supplied to the user terminal 2 without any risk of being able to go back to 1 st NN.
  • CNN Identity can include only two or three layers of convolution (although we will not be limited to any architecture).
  • each CNN Identity aims to approximate the identity as well as possible, possibly with a little noise.
  • the Applicant has observed that this works particularly well (see the document A Protection against the Extraction of Neural Network Models, Hervé Chabanne, Vincent DeLite, Linda Guiga, https: // arxiv. Ors / pdf / 2005.12782 df), and in addition it It is also possible to add even more variability in the set of CNN Identity by adding various noises.
  • Tests were carried out by taking as 1 e NN a ReLU NN with three hidden layers of the fully connected network (FCN) type, the hidden layers having respectively 512, 512 and 32 input channels, this FCN being used for handwritten digit recognition (classification of input images of any size).
  • This 1st NN can be trained for the task based learning MNIST (Mixed National Institute of Standards and Technology) and then shows a correct classification rate of 97.9%
  • the CNN Identity evoked before input size 16x16 (256 channels) can be trained on 10,000 random frames, and this gives an average absolute error between input and output of 0.0913%.
  • step (a) of construction of the 2 nd NN from models of the 1 st NN and CNN Identity the case applicable implementing the selections mentioned before to determine the architecture of the 2nd NN, and only then we learn the parameters of the 2nd NN based learning 1 e NN (eg based NIST mentioned above ).
  • step (a) of construction of the 2 nd NN from models of the 1 st NN and CNN Identity, the case applicable implementing the selections mentioned before to determine the architecture of the 2nd NN, and only then we learn the parameters of the 2nd NN based learning 1 e NN (eg based NIST mentioned above ).
  • the invention relates to a method of learning a second neural network, implemented by the data processing means 11 of the server 1, again comprising step (a) of construction of the second neural network corresponding to a first neural network into which is inserted at least one convolutional neural network approximating the identity function; then a step (a1) of learning the parameters of the second neural network from a public learning database.
  • this learning method can be used as part of a method of secure use of a first neural network on input data (like the method according to the first aspect), by adding the same step (b) of using the 2 nd NN to said input data (implemented this time by the data processing means 21 of terminal 1), ie the 2 nd NN is applied to the data input and one obtains an output datum which can be supplied to the user of the terminal 2 without any risk of being able to go back to the 1 st NN.
  • the invention relates to a computer program product comprising code instructions for execution (in particular on the data processing means 11, 21 of the server 1 or of the terminal 2) of a method according to the first or the third aspect of the invention of secure use of a first neural network on input data or a method according to the second aspect of the invention for learning from parameters of a second neural network, as well as storage means readable by computer equipment (a memory 12, 22 of the server 1 or of the terminal 2) on which this computer program product is found.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Security & Cryptography (AREA)
  • Computer Hardware Design (AREA)
  • Image Analysis (AREA)
EP21732457.3A 2020-05-18 2021-05-14 Procédés d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée, et d'apprentissage de paramètres d'un deuxième réseau de neurones Pending EP4154189A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR2004945A FR3110268B1 (fr) 2020-05-18 2020-05-18 Procédés d’utilisation sécurisée d’un premier réseau de neurones sur une donnée d’entrée, et d’apprentissage de paramètres d’un deuxième réseau de neurones
PCT/FR2021/050842 WO2021234252A1 (fr) 2020-05-18 2021-05-14 Procédés d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée, et d'apprentissage de paramètres d'un deuxième réseau de neurones

Publications (1)

Publication Number Publication Date
EP4154189A1 true EP4154189A1 (fr) 2023-03-29

Family

ID=72644318

Family Applications (1)

Application Number Title Priority Date Filing Date
EP21732457.3A Pending EP4154189A1 (fr) 2020-05-18 2021-05-14 Procédés d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée, et d'apprentissage de paramètres d'un deuxième réseau de neurones

Country Status (5)

Country Link
US (1) US20230196073A1 (cg-RX-API-DMAC7.html)
EP (1) EP4154189A1 (cg-RX-API-DMAC7.html)
JP (1) JP7521010B2 (cg-RX-API-DMAC7.html)
FR (1) FR3110268B1 (cg-RX-API-DMAC7.html)
WO (1) WO2021234252A1 (cg-RX-API-DMAC7.html)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR3133469B1 (fr) 2022-03-09 2024-11-01 Idemia Identity & Security France Procédé d’utilisation sécurisée d’un premier réseau de neurones sur une donnée d’entrée

Family Cites Families (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11676024B2 (en) * 2016-02-24 2023-06-13 Sri International Low precision neural networks using subband decomposition
US10339445B2 (en) * 2016-10-10 2019-07-02 Gyrfalcon Technology Inc. Implementation of ResNet in a CNN based digital integrated circuit
RU2667879C1 (ru) * 2017-05-30 2018-09-24 Общество с ограниченной ответственностью "Аби Продакшн" Обработка и анализ данных на изображениях компьютерной томографии
US10650929B1 (en) * 2017-06-06 2020-05-12 PathAI, Inc. Systems and methods for training a model to predict survival time for a patient
US11023593B2 (en) * 2017-09-25 2021-06-01 International Business Machines Corporation Protecting cognitive systems from model stealing attacks
US10679129B2 (en) * 2017-09-28 2020-06-09 D5Ai Llc Stochastic categorical autoencoder network
US11315570B2 (en) * 2018-05-02 2022-04-26 Facebook Technologies, Llc Machine learning-based speech-to-text transcription cloud intermediary
KR102198817B1 (ko) * 2018-09-12 2021-01-05 주식회사 석영시스템즈 열·공조 시스템에 대한 수요 반응 결정 모델을 생성하는 방법 및 수요 반응을 수행하기 위한 방법
CN109886210B (zh) 2019-02-25 2022-07-19 百度在线网络技术(北京)有限公司 一种交通图像识别方法、装置、计算机设备和介质
US11227187B1 (en) * 2019-05-23 2022-01-18 Augustus Intelligence Inc. Generating artificial intelligence solutions using raw data and simulated data
US11527319B1 (en) * 2019-09-13 2022-12-13 PathAI, Inc. Systems and methods for frame-based validation
EP4038631B1 (en) * 2019-10-03 2025-07-02 Howmedica Osteonics Corp. Cascade of machine learning models to suggest implant components for use in orthopedic joint repair surgeries

Also Published As

Publication number Publication date
FR3110268B1 (fr) 2022-10-21
US20230196073A1 (en) 2023-06-22
JP2023526809A (ja) 2023-06-23
FR3110268A1 (fr) 2021-11-19
JP7521010B2 (ja) 2024-07-23
WO2021234252A1 (fr) 2021-11-25

Similar Documents

Publication Publication Date Title
EP3707676B1 (fr) Procédé d'estimation de pose d'une caméra dans le référentiel d'une scène tridimensionnelle, dispositif, système de réalite augmentée et programme d'ordinateur associé
FR3079329A1 (fr) Procedes d'apprentissage de parametres d'un reseau de neurones a convolution, et de classification d'une donnee d'entree
FR2884008A1 (fr) Systeme et procede de localisation de points d'interet dans une image d'objet mettant en oeuvre un reseau de neurones
FR3087558A1 (fr) Procede d'extraction de caracteristiques d'une empreinte digitale representee par une image d'entree
FR3069357A1 (fr) Systeme d'apprentissage machine pour diverses applications informatiques
EP3736744A1 (fr) Procédé de classification sécurisée d'une donnée d'entrée au moyen d'un réseau de neurones à convolution
EP2407921A1 (fr) Procédé de compression de données pour un classifieur
EP3633548B1 (fr) Procédé de classification sécurisée d'une donnée d'entrée au moyen d'un réseau de neurones à convolution
WO2019137957A1 (fr) Ensemble local d'affichage numérique et réseau de diffusion de contenus numériques comportant de tels ensembles
CA2727747A1 (fr) Dispositif d'aide a la reconnaissance d'images ameliore
FR3112413A1 (fr) Dispositif électronique et procédé de traitement de données à base de réseaux génératifs inversibles, système électronique de détection et programme d’ordinateur associés
EP3966739B1 (fr) Procédé d'analyse automatique d'images pour reconnaître automatiquement au moins une caractéristique rare
WO2021234252A1 (fr) Procédés d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée, et d'apprentissage de paramètres d'un deuxième réseau de neurones
FR3126529A1 (fr) Procédé de mise en relation d’une image candidate avec une image de référence.
EP3825915A1 (fr) Procede de classification d'une empreinte biometrique representee par une image d'entree
EP4242922B1 (fr) Procédé d'utilisation sécurisée d'un premier réseau de neurones sur une donnée d'entrée
EP2880557B1 (fr) Procédé de traitement de données de connexion d'une plateforme d'un site internet
WO2025141279A1 (fr) Procédé d'apprentissage continu pour réseau de neurones de biométrie par comparaison faciale
EP4163866A1 (fr) Procédé, programme d'ordinateur et dispositif de traitement d'images par extraction de composante(s) principale(s) d'une représentation d'images
FR3133937A1 (fr) Procédé et dispositif d’entraînement d’une structure de neurones artificiels et programme d’ordinateur associé
WO2004021265A2 (fr) SystEme associatif flou de description d'objets multimEdia
EP4607479A1 (fr) Méthode et système d'encodage d'une liste de minuties d'un dactylogramme
FR3138223A1 (fr) Procédé, dispositif et programme d’ordinateur de contrôle d’accès à des services numériques
Tanielian Generative Adversarial Networks: theory and practice
FR3153436A1 (fr) Procédé d’authentification ou d’identification multi-biométrique.

Legal Events

Date Code Title Description
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: UNKNOWN

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE

PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE

17P Request for examination filed

Effective date: 20221216

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

DAV Request for validation of the european patent (deleted)
DAX Request for extension of the european patent (deleted)
P01 Opt-out of the competence of the unified patent court (upc) registered

Effective date: 20230919

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: IDEMIA PUBLIC SECURITY FRANCE

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: EXAMINATION IS IN PROGRESS

17Q First examination report despatched

Effective date: 20260708