Home>

by means NUMPY. . I need an array of type a= [[b, c], [d, e] ... [x, z]] In order for in it, both positive and negative numbers.

  • Answer # 1
    Import NUMPY AS NP
    np.random.uniform (Low= -100, High= 100, Size= (4, 2))
    # output
    Array ([
       [-81.458, -81.53563001],
       [14.38820047, 79.58667144],
       [10.55146126, -73.9337667],
       [12.82037855, 71.93397549]])
    

    either so

    np.random.randint (-100, 100, (3, 2))
    Array ([
       [68, -63],
       [41, 52],
       [54, 85]])
    

    To go nuts, all the documentation lasts, did not see this function ... thanks

    Resoxape2021-04-07 18:36:49

    And in what section of the documentation this method is Uniform? I want to look about him yet. Is it creating an array or change?

    Resoxape2021-04-07 18:40:38

    @Resoxape, numpy.org/doc/stable/reference/random/generated/..., there is a site search. Creature.

    entithat2021-04-07 18:41:49
  • Answer # 2

    in> ThisThe article was a moment with a random initialization of the weights of the neural network:

    rows= 10
    COLS= 2.
    Array= 2 * numpy.random.random ((Rows, Cols)) -1
    Print (Array)
    

    Possible conclusion:

    Array ([[-0.90498638, 0.49846144],
           [0.65269211, -0.83121061],
           [0.03048237, 0.27927699],
           [0.72069673, -0.81232947],
           [0.27523583, 0.95059816],
           [0.02586882, 0.96637237],
           [0.88832145, -0.14137144],
           [-0.39758786, 0.43305011],
           [0.32713238, -0.0213639],
           [-0.86635081, -0.63139956]])
    

    Also an important quote from it:

    Note that it is initialized randomly, and average The value is zero. This costs a rather complicated theory. While Just take it as a recommendation.