It's Not What Machines Can Learn, It's What We Cannot Teach

02/21/2020
by   Gal Yehuda, et al.
0

Can deep neural networks learn to solve any task, and in particular problems of high complexity? This question attracts a lot of interest, with recent works tackling computationally hard tasks such as the traveling salesman problem and satisfiability. In this work we offer a different perspective on this question. Given the common assumption that NP≠coNP we prove that any polynomial-time sample generator for an NP-hard problem samples, in fact, from an easier sub-problem. We empirically explore a case study, Conjunctive Query Containment, and show how common data generation techniques generate biased datasets that lead practitioners to over-estimate model accuracy. Our results suggest that machine learning approaches that require training on a dense uniform sampling from the target distribution cannot be used to solve computationally hard problems, the reason being the difficulty of generating sufficiently large and unbiased training sets.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
09/07/2022

The (Un)Scalability of Heuristic Approximators for NP-Hard Search Problems

The A* algorithm is commonly used to solve NP-hard combinatorial optimiz...
research
01/22/2020

A Pseudopolynomial Algorithm to Minimize Maximum Lateness on Multiple Related Machines

In this paper, we will find a pseudopolynomial algorithm to solve Qm || ...
research
03/29/2021

The Inverse Problems of some Mathematical Programming Problems

The non-convex quadratic orogramming problem and the non-monotone linear...
research
06/22/2016

Étude de Problèmes d'Optimisation Combinatoire à Multiples Composantes Interdépendantes

This extended abstract presents an overview on NP-hard optimization prob...
research
02/19/2021

Training Neural Networks is ER-complete

Given a neural network, training data, and a threshold, it was known tha...
research
07/23/2019

ISEA: Image Steganalysis using Evolutionary Algorithms

NP-hard problems always have been attracting scientists' attentions, and...
research
08/02/2020

Complexity continuum within Ising formulation of NP problems

A promising approach to achieve computational supremacy over the classic...

Please sign up or login with your details

Forgot password? Click here to reset