System and computer-based method for simulating a human-like control behaviour in an environmental context

A computer based method for performing machine learning related to a bi-directional compression of high dimensional data by compressing the data into a lower dimensional map, the method comprising:

• receiving data with a first dimension greater than one captured and transmitted by at least one sensor of a vehicle;
• bi-directionally compressing the data or attributes derived from the data into compressed data with a second dimension lower than the first dimension;
• and
• using machine learning based on the compressed data, generating and transmitting a signal including data corresponding to a state of the vehicle to a component of the vehicle,
• wherein, for two metric spaces X=(x,μx) and Y=(y,μy) with x a first space having a dimension that is greater than or equal to a dimension of a second space y,μx a first metric on the first space x,μy a second metric on the second space y, the compression being performed by mapping pairs of points formed between a first set of n points P belonging to the first space x and a second set of n points Q belonging to the second space y, each point of the second set of n points Q being paired with a point of the first set of points P through the following equation:
• Q=argmint∈ynd(A1(μx(P)),A2(μy(t))) (Equation 1)
• where μx(P) is a first symmetric distance matrix between points in P, μx(Q) is a second symmetric distance matrix between points in Q, and d(.,.) is a symmetric distance function on the space of square matrices, while A1 and A2 are any two functions holding the following properties:
• A:
• R+→[0, 1],
• A(0)=1,
• limx→∞A(z)=0, and
• A(z1)>A(z2) if and only if z1<z2.