In this paper, we prove that most of the boolean functions, $f : \{-1,1\}^n \rightarrow \{-1,1\}$ satisfy the Fourier Entropy Influence (FEI) Conjecture due to Friedgut and Kalai (Proc. AMS'96). The conjecture says that the Entropy of a boolean function is at most a constant times the Influence of the function. The conjecture has been proven for families of functions of smaller sizes. O'donnell, Wright and Zhou (ICALP'11) verified the conjecture for the family of symmetric functions, whose size is $2^{n+1}$. They are in fact able to prove the conjecture for the family of $d$-part symmetric functions for constant $d$, the size of whose is $2^{O(n^d)}$. Also it is known that the conjecture is true for a large fraction of polynomial sized DNFs (COLT'10). Using elementary methods we prove that a random function with high probability satisfies the conjecture with the constant as $(2 + \delta)$, for any constant $\delta > 0$.
[1]
A. Bernasconi,et al.
On the Fourier Analysis of Boolean Functions
,
2007
.
[2]
Ryan O'Donnell,et al.
The Fourier Entropy-Influence Conjecture for Certain Classes of Boolean Functions
,
2011,
ICALP.
[3]
Yishay Mansour,et al.
Learning Boolean Functions via the Fourier Transform
,
1994
.
[4]
G. Kalai,et al.
Every monotone graph property has a sharp threshold
,
1996
.
[5]
Thomas M. Cover,et al.
Elements of Information Theory
,
2005
.
[6]
Andrew Wan,et al.
Mansour's Conjecture is True for Random DNF Formulas
,
2010,
COLT.
[7]
Elchanan Mossel,et al.
A note on the Entropy/Influence conjecture
,
2012,
Discret. Math..