giorgiop/loss-correction

Different estimated T in figure 1?

Closed this issue · 4 comments

Hi, @giorgiop. In your code, the process of estimating backward T and forward T is just the same, so how could you get different accurracy of backward T and forward T with different noise rate in your paper?
estimate t

It is true that the estimated matrix T is the same. But it is then used by the two different algorithms for loss correction, so the results are different.

I have misunderstood the figure.
T is known (solid line), T is estimated (dotted line).
@giorgiop ,Is that right?

Correct!

Thank you!