Method for configuring a sensor detection device, and corresponding computer program and adaptive device

A method for configuring a detection device for detecting a situation from a set of situations wherein a physical system observed by at least one sensor is liable to be, comprising the following steps:

• reception of a sequence of observation data for the physical system, referred to as a learning sequence, supplied by the sensor and corresponding to a given situation of the physical system,
• determination, from the learning sequence, of parameters of a hidden-state Markov statistical model recorded in storage means of the detection device and relating to the given situation, by prior initialisation of these parameters, and then updating of these initialised parameters,
• wherein the prior initialisation comprises the following steps:
• the statistical model in question comprising a given number of hidden states, determination of a plurality of probability distributions from the learning sequence, by dividing the sequence into sub-sequences and allocating to each sub-sequence a probability distribution that models it statistically, the number of given probability distributions being greater than the number of hidden states of the statistical model in question,
• distribution of the probability distributions determined between the various hidden states of the statistical model in question,
• determination, for each hidden state of the statistical model in question and from the probability distributions allocated to this hidden state, of a single probability distribution representing this hidden state, and
• initialisation of the parameters of the statistical model in question from the determined representative probability distributions,
• characterised in that, the statistical model in question further comprising impossible transition constraints between certain hidden states, the distribution of the probability distributions determined between the various hidden states of the statistical model in question is done by global optimisation of a function of adaptation of these probability distributions to the various hidden states and to the impossible transition constraints, and the method further comprises a step of configuring the detection device so that the statistical model in question includes the parameters determined by said prior initialisation and then said updating.