The Circuit Complexity of Inference

04/11/2019
by   Ankur Moitra, et al.
0

Belief propagation is one of the foundations of probabilistic and causal reasoning. In this paper, we study the circuit complexity of some of the various tasks it performs. Specifically, in the broadcast tree model (which has important applications to phylogenetic reconstruction and close connections to community detection), we show the following: (1) No AC^0 circuit can guess the label of the root with positive advantage over random guessing, independent of the depth for any non-trivial choice of parameters. (2) There is a TC^0 circuit that competes with the Bayes optimal predictor in some range of parameters above the Kesten-Stigum bound. (3) There is a 16 label broadcast tree model in which it is possible to accurately guess the label of the root, but beating random guessing is NC^1-hard. Our work yields a simple and natural generative model where large depth really is necessary for performing various types of inference, that have intriguing parallels with phase transitions from statistical physics.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
09/14/2021

Reconstruction on Trees and Low-Degree Polynomials

The study of Markov processes and broadcasting on trees has deep connect...
research
06/28/2018

Typology of phase transitions in Bayesian inference problems

Many inference problems, notably the stochastic block model (SBM) that g...
research
01/25/2021

Circuit Complexity From Supersymmetric Quantum Field Theory With Morse Function

Computation of circuit complexity has gained much attention in the Theor...
research
05/24/2019

Accuracy-Memory Tradeoffs and Phase Transitions in Belief Propagation

The analysis of Belief Propagation and other algorithms for the reconst...
research
09/22/2017

Quantified Derandomization of Linear Threshold Circuits

One of the prominent current challenges in complexity theory is the atte...
research
06/18/2022

Parameter Identifiability of a Multitype Pure-Birth Model of Speciation

Diversification models describe the random growth of evolutionary trees,...
research
08/01/2022

Experimental validation of the Kibble-Zurek Mechanism on a Digital Quantum Computer

The Kibble-Zurek mechanism (KZM) captures the essential physics of noneq...

Please sign up or login with your details

Forgot password? Click here to reset