Homepage
About membership
Sign inGet started
Homepage

Kaustubh Sable

Kaustubh Sable
34 Following22 Followers
  • Profile
  • Claps
  • Highlights
Highlighted by Kaustubh Sable
See more
  • From Batch Normalization — What the hey? by Karl N.

    …ulation versus the sample/batch variance is that σ is normalized by m and s is normalized by (m-1). Conceptually, the reason why we use (m-1) is that you’re also estimating the mean (in the form of x bar) to be centered, you’ll have to expect that the spread of your points (i.e., variance) will be a bit larger. In this case, it’s going to be precisely m/(m-1) larger. So, when we assign

Claps from Kaustubh Sable
See more
  • Mathematical Optimization: ‘simplicity is all you need’
    Mathematical Optimization: ‘simplicity is all you need’
    Animesh Karnewar
  • Advice From A 19 Year Old Girl & Software Developer
    Advice From A 19 Year Old Girl & Software Developer
    Lydia Hallie
  • Thinking in the Hough Space: The Doing
    Thinking in the Hough Space: The Doing
    Vignesh Ungrapalli