average case lower bounds for threshold circuits
play

Average case lower bounds for threshold circuits Ruiwen Chen, Rahul - PowerPoint PPT Presentation

Jan 16, 2024 •219 likes •1.18k views

Average case lower bounds for threshold circuits Ruiwen Chen, Rahul Santhanam and Srikanth Srinivasan University of Oxford and Department of Mathematics, IIT Bombay CCC 2016 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 1 / 21

  1. Average case lower bounds for threshold circuits Ruiwen Chen, Rahul Santhanam and Srikanth Srinivasan University of Oxford and Department of Mathematics, IIT Bombay CCC 2016 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 1 / 21

  2. Boolean Circuits Circuit computing function f : { 0 , 1 } n → { 0 , 1 } . g 1 Computation proceeds through “simple” g 3 operations. g 2 x 1 x 2 x 3 x 4 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 2 / 21

  3. Boolean Circuits Circuit computing function f : { 0 , 1 } n → { 0 , 1 } . g 1 Computation proceeds through “simple” g 3 operations. g 2 g i ∈ “basic” operations. x 1 x 2 x 3 x 4 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 2 / 21

  4. Boolean Circuits Circuit computing function f : { 0 , 1 } n → { 0 , 1 } . g 1 Computation proceeds through “simple” g 3 operations. g 2 g i ∈ “basic” operations. Designated output gate x 1 x 2 x 3 x 4 computes function f . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 2 / 21

  5. Boolean Circuits Size s of the circuit: time g 1 taken by algorithm. g 3 g 2 depth = 3 x 1 x 2 x 3 x 4 # wires = 8 , # gates = 3 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 3 / 21

  6. Boolean Circuits Size s of the circuit: time g 1 taken by algorithm. Could be # edges/wires or # gates. g 3 g 2 depth = 3 x 1 x 2 x 3 x 4 # wires = 8 , # gates = 3 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 3 / 21

  7. Boolean Circuits Size s of the circuit: time g 1 taken by algorithm. Could be # edges/wires or # gates. g 3 # wires ≤ ( n + # gates) · g 2 depth = 3 # gates. x 1 x 2 x 3 x 4 # wires = 8 , # gates = 3 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 3 / 21

  8. Boolean Circuits Size s of the circuit: time g 1 taken by algorithm. Could be # edges/wires or # gates. g 3 # wires ≤ ( n + # gates) · g 2 depth = 3 # gates. Depth d of the circuit: parallelism of the x 1 x 2 x 3 x 4 algorithm. # wires = 8 , # gates = 3 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 3 / 21

  9. Boolean Circuits Size s of the circuit: time g 1 taken by algorithm. Could be # edges/wires or # gates. g 3 # wires ≤ ( n + # gates) · g 2 depth = 3 # gates. Depth d of the circuit: parallelism of the x 1 x 2 x 3 x 4 algorithm. # wires = 8 , # gates = 3 s = s ( n ) , d = O (1) . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 3 / 21

  10. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  11. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  12. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . AND function: AND( x ) = � � i x i ≥ n � . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  13. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . AND function: AND( x ) = � � i x i ≥ n � . MAJ function: MAJ( x ) = � � i x i ≥ n/ 2 � . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  14. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . AND function: AND( x ) = � � i x i ≥ n � . MAJ function: MAJ( x ) = � � i x i ≥ n/ 2 � . i 2 i ( x i − y i ) ≥ 0 � . GEQ function: GEQ( x, y ) = � � Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  15. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . AND function: AND( x ) = � � i x i ≥ n � . MAJ function: MAJ( x ) = � � i x i ≥ n/ 2 � . i 2 i ( x i − y i ) ≥ 0 � . GEQ function: GEQ( x, y ) = � � TC 0 g ( s, d ) : threshold circuits with s gates and depth d . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  16. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . AND function: AND( x ) = � � i x i ≥ n � . MAJ function: MAJ( x ) = � � i x i ≥ n/ 2 � . i 2 i ( x i − y i ) ≥ 0 � . GEQ function: GEQ( x, y ) = � � TC 0 g ( s, d ) : threshold circuits with s gates and depth d . TC 0 w ( s, d ) : threshold circuits with s wires and depth d . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  17. Threshold circuits A threshold operation: g ( x ) = 1 iff � i w i x i ≥ θ for w i , θ ∈ R . Some examples: OR function: OR( x ) = 1 iff � i x i ≥ 1 . AND function: AND( x ) = � � i x i ≥ n � . MAJ function: MAJ( x ) = � � i x i ≥ n/ 2 � . i 2 i ( x i − y i ) ≥ 0 � . GEQ function: GEQ( x, y ) = � � TC 0 g ( s, d ) : threshold circuits with s gates and depth d . TC 0 w ( s, d ) : threshold circuits with s wires and depth d . Generalize AC 0 circuits made up of AND and OR gates. Average case bounds for TC 0 Chen, Santhanam, S. May 2016 4 / 21

  18. The power of threshold circuits f = PARITY( x 1 , . . . , x n ) = x 1 ⊕ x 2 ⊕ · · · ⊕ x n . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 5 / 21

  19. The power of threshold circuits f = PARITY( x 1 , . . . , x n ) = x 1 ⊕ x 2 ⊕ · · · ⊕ x n . d ≥ 2 : f ∈ TC 0 g ( dn 1 / ( d − 1) , d ) (Siu-Roychowdhury-Kailath 1991) Average case bounds for TC 0 Chen, Santhanam, S. May 2016 5 / 21

  20. The power of threshold circuits f = PARITY( x 1 , . . . , x n ) = x 1 ⊕ x 2 ⊕ · · · ⊕ x n . d ≥ 2 : f ∈ TC 0 g ( dn 1 / ( d − 1) , d ) (Siu-Roychowdhury-Kailath 1991) w ( n 1+ ε d , d ) (Beame-Brisson-Ladner, Paturi-Saks 1991) f ∈ TC 0 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 5 / 21

  21. The power of threshold circuits f = PARITY( x 1 , . . . , x n ) = x 1 ⊕ x 2 ⊕ · · · ⊕ x n . d ≥ 2 : f ∈ TC 0 g ( dn 1 / ( d − 1) , d ) (Siu-Roychowdhury-Kailath 1991) w ( n 1+ ε d , d ) (Beame-Brisson-Ladner, Paturi-Saks 1991) f ∈ TC 0 Compare with: PARITY does not have AC 0 circuits of subexponential size (H˚ astad 1986). Average case bounds for TC 0 Chen, Santhanam, S. May 2016 5 / 21

  22. Circuit lower bounds Problem: Find explicit family of functions (say in NP) that have no TC 0 circuits of poly ( n ) size. Average case bounds for TC 0 Chen, Santhanam, S. May 2016 6 / 21

  23. Circuit lower bounds Problem: Find explicit family of functions (say in NP) that have no TC 0 circuits of poly ( n ) size. Even open for depth 2 . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 6 / 21

  24. Work on threshold circuits Hajnal Maass Pudl´ ak Turan Szegedy 1987 (Polynomial Approximations) Paturi Saks 1991, Siu Roychowdhury Kailath 1992; Beigel 1994; Aspnes Beigel Furst Rudich 1994, Podolskii 2012 (Combinatorial restrictions) Impagliazzo Paturi Saks 1991 (Communication complexity) Goldmann Hastad Razborov 1992; Nisan 1992; Hansen Miltersen 2004; Chattopadhyay Hansen 2005; Lovett, S. 2012 (Analytic techniques) Gopalan Servedio 2010 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 7 / 21

  25. State-of-the-art lower bounds (Impagliazzo-Paturi-Saks 1991) PARITY not in TC 0 g ( n 1 / 2( d − 1) , d ) and w ( n 1+ ε d , d ) . TC 0 Average case bounds for TC 0 Chen, Santhanam, S. May 2016 8 / 21

  26. State-of-the-art lower bounds (Impagliazzo-Paturi-Saks 1991) PARITY not in TC 0 g ( n 1 / 2( d − 1) , d ) and w ( n 1+ ε d , d ) . TC 0 (Kane-Williams 2015) Explicit functions not in TC 0 g ( n 1 . 5 − o (1) , 2) and TC 0 w ( n 2 . 5 − o (1) , 2) . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 8 / 21

  27. State-of-the-art lower bounds (Impagliazzo-Paturi-Saks 1991) PARITY not in TC 0 g ( n 1 / 2( d − 1) , d ) and w ( n 1+ ε d , d ) . TC 0 (Kane-Williams 2015) Explicit functions not in TC 0 g ( n 1 . 5 − o (1) , 2) and TC 0 w ( n 2 . 5 − o (1) , 2) . Also extends to a special case of depth- 3 . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 8 / 21

  28. Average case lower bounds Want to show a function f : { 0 , 1 } n → { 0 , 1 } hard on average . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 9 / 21

  29. Average case lower bounds Want to show a function f : { 0 , 1 } n → { 0 , 1 } hard on average . Trivial to compute f on half the inputs. Average case bounds for TC 0 Chen, Santhanam, S. May 2016 9 / 21

  30. Average case lower bounds Want to show a function f : { 0 , 1 } n → { 0 , 1 } hard on average . Trivial to compute f on half the inputs. f has ε -correlation with ckt C if x [ C ( x ) = f ( x )] − 1 Corr( C, f ) := Pr 2 ≤ ε. Average case bounds for TC 0 Chen, Santhanam, S. May 2016 9 / 21

  31. Average case lower bounds Want to show a function f : { 0 , 1 } n → { 0 , 1 } hard on average . Trivial to compute f on half the inputs. f has ε -correlation with ckt C if x [ C ( x ) = f ( x )] − 1 Corr( C, f ) := Pr 2 ≤ ε. Want to show that f hard on average against TC 0 ( s, d ) . Average case bounds for TC 0 Chen, Santhanam, S. May 2016 9 / 21

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Average - case Lower Bounds for Approximate Near - Neighbor fs om Isoperimetric Inequalities

Average - case Lower Bounds for Approximate Near - Neighbor fs om Isoperimetric Inequalities

Simple Average - case Lower Bounds for Approximate Near - Neighbor fs om Isoperimetric Inequalities Yitong Yin Nanjing University Nearest Neighbor Search ( NNS ) metric space ( X, dist) query x X database access y = ( y 1 , y 2 , . . . , y

384 views • 26 slides
Lower Bounds Best-, Average-, and Worst-Case Time Complexity Ex. Insertion sort

Lower Bounds Best-, Average-, and Worst-Case Time Complexity Ex. Insertion sort

Lower Bounds Best-, Average-, and Worst-Case Time Complexity Ex. Insertion sort of x 1 , x 2 , , x n . For i = 2, 3, , n , insert x i into x 1 , x 2 , , x i 1 such that these i data items are sorted.

509 views • 31 slides
Heapsort In the last class Mergesort Worst Case Analysis of Mergesort Lower Bounds

Heapsort In the last class Mergesort Worst Case Analysis of Mergesort Lower Bounds

Algorithm : Design & Analysis [6] Heapsort In the last class Mergesort Worst Case Analysis of Mergesort Lower Bounds for Sorting by Comparison of Keys Worst Case Average Behavior Heapsort Heap Structure and Patial

658 views • 31 slides
Exponential lower bounds for hom. depth-5 circuits over finite fields Mrinal Kumar Ramprasad

Exponential lower bounds for hom. depth-5 circuits over finite fields Mrinal Kumar Ramprasad

Exponential lower bounds for hom. depth-5 circuits over finite fields Mrinal Kumar Ramprasad Saptharishi TIFR, Mumbai CCC 2017 Riga Rutgers Harvard Algebraic Circuits f ( x 1 , x 2 , x 3 ) = 2 x 2 1 + 2 x 1 x 2 + 2 x 1 x 3 + 2 x 2 x 3 =

1.56k views • 115 slides
Circuit Lower-bounds Lecture 24 Weak circuits are indeed weak 1 Circuit Lower-bounds 2

Circuit Lower-bounds Lecture 24 Weak circuits are indeed weak 1 Circuit Lower-bounds 2

Circuit Lower-bounds Lecture 24 Weak circuits are indeed weak 1 Circuit Lower-bounds 2 Circuit Lower-bounds Today: 2 Circuit Lower-bounds Today: PARITY AC 0 2 Circuit Lower-bounds Today: PARITY AC 0 Two different proofs! (Latter

1.59k views • 137 slides
Stronger Connections Between Circuit Analysis and Circuit Lower Bounds, via PCPs of Proximity

Stronger Connections Between Circuit Analysis and Circuit Lower Bounds, via PCPs of Proximity

Stronger Connections Between Circuit Analysis and Circuit Lower Bounds, via PCPs of Proximity Lijie Chen Ryan Williams Context: The Algorithmic Method for Proving Circuit Lower Bounds Proving limitations on non-uniform circuits is extremely

796 views • 31 slides
Functional lower bounds for arithmetic circuits and connections to boolean circuit complexity

Functional lower bounds for arithmetic circuits and connections to boolean circuit complexity

Functional lower bounds for arithmetic circuits and connections to boolean circuit complexity Michael Forbes Mrinal Kumar Ramprasad Saptharishi Princeton University Rutgers University T el Aviv University Computational Complexity

1.35k views • 107 slides
Lower bounds for average values of Arakelov-Green functions and global applications Matt Baker

Lower bounds for average values of Arakelov-Green functions and global applications Matt Baker

Motivation: The work of HindrySilverman Equidistribution and Non-Equidistribution on Elliptic Curves Arakelov-Green functions on metrized graphs Arithmetic dynamics Lower bounds for average values of Arakelov-Green functions and global

1.46k views • 102 slides
A Superpolynomial Lower Bound for Clique Function Circuits with at most 1 6 loglog n Negation

A Superpolynomial Lower Bound for Clique Function Circuits with at most 1 6 loglog n Negation

Main Results Hawks Eye View Extending Montone Lower Bounds to Limited Negations Monotone Lower bound for Clique function using Bottleneck counting Connecting the two worlds Thank You A Superpolynomial Lower Bound for Clique Function

1.03k views • 71 slides
9. Sorting III Lower bounds for the comparison based sorting, radix- and bucket-sort 248 9.1

9. Sorting III Lower bounds for the comparison based sorting, radix- and bucket-sort 248 9.1

9. Sorting III Lower bounds for the comparison based sorting, radix- and bucket-sort 248 9.1 Lower bounds for comparison based sorting [Ottman/Widmayer, Kap. 2.8, Cormen et al, Kap. 8.1] 249 Lower bound for sorting Up to here: worst case

600 views • 55 slides
Polynomial threshold functions and Boolean threshold circuits Kristoffer Arnsfelt Hansen 1

Polynomial threshold functions and Boolean threshold circuits Kristoffer Arnsfelt Hansen 1

Polynomial threshold functions and Boolean threshold circuits Kristoffer Arnsfelt Hansen 1 Vladimir V. Podolskii 2 1 Aarhus University 2 Steklov Mathematical Institute MFCS 2013 1 / 27 Boolean Threshold Functions Boolean function f : { a , b } n

927 views • 44 slides
Conspiracies between Learning Algorithms, Lower Bounds, and Pseudorandomness Igor Carboni

Conspiracies between Learning Algorithms, Lower Bounds, and Pseudorandomness Igor Carboni

Conspiracies between Learning Algorithms, Lower Bounds, and Pseudorandomness Igor Carboni Oliveira University of Oxford Joint work with Rahul Santhanam (Oxford) Context Minor algorithmic improvements imply lower bounds (Williams, 2010). NEXP

696 views • 30 slides
Lower and upper estimates on the excitation threshold for DNLS lattices J. Cuevas, N.I.

Lower and upper estimates on the excitation threshold for DNLS lattices J. Cuevas, N.I.

Definition of the excitation threshold Upper and lower estimates for the excitation threshold Figures from numerical studies Discussion Lower and upper estimates on the excitation threshold for DNLS lattices J. Cuevas, N.I. Karachalios and F.

246 views • 22 slides
Lower bounds for parameterized problems Dniel Marx 1 1 Institute for Computer Science and

Lower bounds for parameterized problems Dniel Marx 1 1 Institute for Computer Science and

Lower bounds for parameterized problems Dniel Marx 1 1 Institute for Computer Science and Control, Hungarian Academy of Sciences (MTA SZTAKI) Budapest, Hungary ICERM Providence, RI April 26, 2014 1 Lower bounds So far we have seen positive

1.14k views • 94 slides
Dispersers and Circuit Lower Bounds Alexander Golovnev New York University ITCS 2016 p i x n

Dispersers and Circuit Lower Bounds Alexander Golovnev New York University ITCS 2016 p i x n

Dispersers and Circuit Lower Bounds Alexander Golovnev New York University ITCS 2016 p i x n [Hs89] 3 f p i x o n [PSZ97] 3 f depth- 3 circuits x 7 x 5 x 4 x 4 x 7 x 5 x 4 x 7 x 5 depth: 3, bottom fan-in: unbounded x 7 x 7 x 5 x 7 x 5

107 views • 8 slides
Lower Bounds on the Probability of a Finite Union of Events Jun Yang (joint work with Fady

Lower Bounds on the Probability of a Finite Union of Events Jun Yang (joint work with Fady

Lower Bounds on the Probability of a Finite Union of Events Jun Yang (joint work with Fady Alajaji and Glen Takahara) Department of Mathematics and Statistics, Queens University, Kingston, Canada ISIT 2014 Lower Bounds on P ( N Jun

228 views • 19 slides
Scalable and Robust Management of Dynamic Graph Data Alan G. Labouseur Paul W. Olsen Jr.

Scalable and Robust Management of Dynamic Graph Data Alan G. Labouseur Paul W. Olsen Jr.

BD 3 2013 Scalable and Robust Management of Dynamic Graph Data Alan G. Labouseur Paul W. Olsen Jr. Jeong-Hyon Hwang {alan, polsen, jhh}@cs.albany.edu Sunday, September 22, 2013 Large, Dynamic Networks BD 3 2013 2 Sunday,

846 views • 57 slides
A graphical model for sequential teams Aditya Mahajan and Sekhar Tatikonda Dept of Electrical

A graphical model for sequential teams Aditya Mahajan and Sekhar Tatikonda Dept of Electrical

A graphical model for sequential teams Aditya Mahajan and Sekhar Tatikonda Dept of Electrical Engineering Yale University Presented at: ConCom Workshop, June 27, 2009 A glimpse of the result Structural results in sequential teams Example:

1.05k views • 57 slides
Computing Parameters of Sequence-based Dynamic Graphs Ralf Klasing LaBRI, CNRS, University of

Computing Parameters of Sequence-based Dynamic Graphs Ralf Klasing LaBRI, CNRS, University of

Computing Parameters of Sequence-based Dynamic Graphs Ralf Klasing LaBRI, CNRS, University of Bordeaux, France **This is a joint work with Arnaud Casteigts, Yessin M. Neggaz, and Joseph G. Peters. Dynamic Networks Highly dynamic networks? 1

1.73k views • 128 slides
e ^ ga . -a4t. .-.\)Aa^... -1:44^. _-4A,1-^* a.ee*+a^r**u . 'l:5 Li2N Figure 6.

e ^ ga . -a4t. .-.\)Aa^... -1:44^. _-4A,1-^* a.ee*+a^r**u . 'l:5 Li2N Figure 6.

" : t q . " . - - i *'l-i',o. $ 1 I 1.''.,;,''":'{ ..,,,"i t. i i':''n u" : * -.i.-'- "t $i p, r'*..u,,,...il-u 1*" F' j 1 1 . 2 i I n R ) . t i n 2l 4 t ( f r 2 " Figure 2'

365 views • 5 slides
Practical validation of several fault attacks against the Miller algorithm Nadia El Mrabet 1 ,

Practical validation of several fault attacks against the Miller algorithm Nadia El Mrabet 1 ,

Practical validation of several fault attacks against the Miller algorithm Nadia El Mrabet 1 , Jacques Fournier 2 , Louis Goubin 3 , Ronan Lashermes 2 , 3 , Marie Paindavoine 4 , 5 . 1 - LIASD, Paris 8, France. 2 - CEA Tech, DPACA/LSAS, Gardanne,

446 views • 22 slides
Elliptic gamma functions, gerbes and triptic curves Giovanni Felder, ETH Zurich Paris, 18

Elliptic gamma functions, gerbes and triptic curves Giovanni Felder, ETH Zurich Paris, 18

Elliptic gamma functions, gerbes and triptic curves Giovanni Felder, ETH Zurich Paris, 18 January 2007 1 Table of contents 0. Introduction 1. Two periods: Jacobis infinite products, elliptic curves, SL 2 ( Z ) 2. Three periods:

510 views • 18 slides
Today Perceptron. Today Perceptron. Support Vector Machine. Labelled points with x 1 ,..., x n

Today Perceptron. Today Perceptron. Support Vector Machine. Labelled points with x 1 ,..., x n

Today Perceptron. Today Perceptron. Support Vector Machine. Labelled points with x 1 ,..., x n . + + ++ Labelled points with x 1 ,..., x n . Hyperplane separator. + + ++ Labelled points

2.36k views • 148 slides
The heteroscedastic odd log-logistic generalized gamma regression model for censored data F

The heteroscedastic odd log-logistic generalized gamma regression model for censored data F

The heteroscedastic odd log-logistic generalized gamma regression model for censored data F abio Prataviera Edwin M. M. Ortega Gauss M. Cordeiro Altemir da Silva Braga VIII - Encontro dos Alunos da P os Gradua c ao em Estat

865 views • 39 slides

More recommend