Method for improving user authentication performed by a communication device

A communication device belonging to an authentication system, the communication device comprising a local machine learning engine comprising a set of N artificial neural network ANN1, i adapted to process N different types of input signals, the communication device being configured to:

• receive a first set of N input signals S_1(i) of different types provided by the requesting user for authentication purpose;
• determine respectively for each of the N input signals S_1(i) by the N artificial neural networks ANN1, i, N estimations LH(i) of the likelihood that a given input signal is provided by a legitimate user allowed to access to at least a service provided by the communication device;
• determine based on a risk scoring established using the N estimations LH(i) if the requesting user is authenticated as the legitimate user;
• if the requested user is authenticated, determining if at least one likelihood estimation determined for a given input signal S_1(j) is below a predetermined threshold Tc,i, and upon determining that at least one likelihood estimation determined for a given input signal S_1(j) is below a predetermined threshold Tc,i:
• transmit (305) the input signal S_1(j) to a remote server implementing a server machine learning engine adapted to process said N different types of input signals and trained to distinguish input signals provided by a plurality of users of the authentication system, a candidate user U_C different from the requesting user for which the transmitted input signal S_1(j) provides the best matching results being identified, a corresponding input signal S_2(j) associated to the candidate user U_C and of the same type as S_1(j) being provided;
• receive the input signal S_2(j) associated to the candidate user U_C and executing by the local machine learning engine an additional learning phase using the input signal S_2(j) as an input signal that is not associated to the requesting user.

An authentication system comprising a communication device and a remote server comprising means for carrying out the steps of a method for improving user authentication efficiency performed by a communication device belonging to an authentication system, the communication device comprising a local machine learning engine comprising a set of N artificial neural network ANN1, i adapted to process N different types of input signals, the method comprising the following steps:

• receiving a first set of N input signals S_1(i) of different types provided by the requesting user for authentication purpose;
• determining respectively for each of the N input signals S_1(i) by the N artificial neural networks ANN1, i, N estimations LH(i) of the likelihood that a given input signal is provided by a legitimate user allowed to access to at least a service provided by the communication device;
• determining based on a risk scoring established using the N estimations LH(i) if the requesting user is authenticated as the legitimate user;
• if the requested user is authenticated, determining if at least one likelihood estimation determined for a given input signal S_1(j) is below a predetermined threshold Tc,i, and upon determining that at least one likelihood estimation determined for a given input signal S_1(j) is below a predetermined threshold Tc,i:
• transmitting the input signal S_1(j) to a remote server implementing a server machine learning engine adapted to process said N different types of input signals and trained to identify a user U_C, called closest candidate, different from the requesting user and for which the transmitted input signal S_1(j) provides the best matching result, the remote server then providing an input signal S_2(j) associated to the closest candidate U_C and of the same type as S_1(j);
• receiving the input signal S_2(j) associated to the closest candidate U_C and executing by the local machine learning engine an additional learning phase using the input signal S_2(j) as an input signal that is not associated to the requesting user.

A non-transitory computer memory storing a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of a method for improving user authentication efficiency performed by a communication device belonging to an authentication system, the communication device comprising a local machine learning engine comprising a set of N artificial neural network ANN1, i adapted to process N different types of input signals, the instructions comprising instructions to cause the computer to perform the following steps:

• receiving a first set of N input signals S_1(i) of different types provided by the requesting user for authentication purpose;
• determining respectively for each of the N input signals S_1(i) by the N artificial neural networks ANN1, i, N estimations LH(i) of the likelihood that a given input signal is provided by a legitimate user allowed to access to at least a service provided by the communication device;
• determining based on a risk scoring established using the N estimations LH(i) if the requesting user is authenticated as the legitimate user;
• if the requested user is authenticated, determining if at least one likelihood estimation determined for a given input signal S_1(j) is below a predetermined threshold Tc,i, and upon determining that at least one likelihood estimation determined for a given input signal S_1(j) is below a predetermined threshold Tc,i:
• transmitting the input signal S_1(j) to a remote server implementing a server machine learning engine adapted to process said N different types of input signals and trained to identify a user U_C, called closest candidate, different from the requesting user and for which the transmitted input signal S_1(j) provides the best matching result, the remote server then providing an input signal S_2(j) associated to the closest candidate U_C and of the same type as S_1(j);
• receiving the input signal S_2(j) associated to the closest candidate U_C and executing by the local machine learning engine an additional learning phase using the input signal S_2(j) as an input signal that is not associated to the requesting user.