Quantum learning algorithms imply circuit lower bounds

We establish the first general connection between the design of quantum algorithms and circuit lower bounds. Specifically, let $\mathfrak{C}$ be a class of polynomial-size concepts, and suppose that $\mathfrak{C}$ can be PAC-learned with membership queries under the uniform distribution with error $1/2 - \gamma$ by a time $T$ quantum algorithm. We prove that if $\gamma^2 \cdot T \ll 2^n/n$, then $\mathsf{BQE} \nsubseteq \mathfrak{C}$, where $\mathsf{BQE} = \mathsf{BQTIME}[2^{O(n)}]$ is an exponential-time analogue of $\mathsf{BQP}$. This result is optimal in both $\gamma$ and $T$, since it is not hard to learn any class $\mathfrak{C}$ of functions in (classical) time $T = 2^n$ (with no error), or in quantum time $T = \mathsf{poly}(n)$ with error at most $1/2 - \Omega(2^{-n/2})$ via Fourier sampling. In other words, even a marginal improvement on these generic learning algorithms would lead to major consequences in complexity theory. Our proof builds on several works in learning theory, pseudorandomness, and computational complexity, and crucially, on a connection between non-trivial classical learning algorithms and circuit lower bounds established by Oliveira and Santhanam (CCC 2017). Extending their approach to quantum learning algorithms turns out to create significant challenges. To achieve that, we show among other results how pseudorandom generators imply learning-to-lower-bound connections in a generic fashion, construct the first conditional pseudorandom generator secure against uniform quantum computations, and extend the local list-decoding algorithm of Impagliazzo, Jaiswal, Kabanets and Wigderson (SICOMP 2010) to quantum circuits via a delicate analysis. We believe that these contributions are of independent interest and might find other applications.

[1]  Gilles Brassard,et al.  Machine Learning in a Quantum World , 2006, Canadian AI.

[2]  Avi Wigderson,et al.  Randomness vs Time: Derandomization under a Uniform Assumption , 2001, J. Comput. Syst. Sci..

[3]  Russell Impagliazzo,et al.  Learning Algorithms from Natural Proofs , 2016, CCC.

[4]  Peter W. Shor,et al.  Algorithms for quantum computation: discrete logarithms and factoring , 1994, Proceedings 35th Annual Symposium on Foundations of Computer Science.

[5]  Stasys Jukna,et al.  Boolean Function Complexity Advances and Frontiers , 2012, Bull. EATCS.

[6]  Luca Trevisan,et al.  Pseudorandomness and Average-Case Complexity Via Uniform Reductions , 2002, Proceedings 17th IEEE Annual Conference on Computational Complexity.

[7]  Moni Naor,et al.  Synthesizers and their application to the parallel construction of pseudo-random functions , 1995, Proceedings of IEEE 36th Annual Foundations of Computer Science.

[8]  Noga Alon,et al.  The Probabilistic Method , 2015, Fundamentals of Ramsey Theory.

[9]  Rocco A. Servedio,et al.  Improved Bounds on Quantum Learning Algorithms , 2004, Quantum Inf. Process..

[10]  Lov K. Grover A fast quantum mechanical algorithm for database search , 1996, STOC '96.

[11]  Ryan O'Donnell,et al.  Efficient quantum tomography , 2015, STOC.

[12]  Lijie Chen,et al.  Strong average-case lower bounds from non-trivial derandomization , 2020, STOC.

[13]  Michael Kharitonov,et al.  Cryptographic hardness of distribution-specific learning , 1993, STOC.

[14]  Rocco A. Servedio,et al.  Equivalences and Separations Between Quantum and Classical Learnability , 2004, SIAM J. Comput..

[15]  Noam Nisan,et al.  Hardness vs Randomness , 1994, J. Comput. Syst. Sci..

[16]  Isaac L. Chuang,et al.  Quantum Computation and Quantum Information (10th Anniversary edition) , 2011 .

[17]  John M. Hitchcock,et al.  Exact Learning Algorithms, Betting Games, and Circuit Lower Bounds , 2013, TOCT.

[18]  Nader H. Bshouty,et al.  Learning DNF over the uniform distribution using a quantum example oracle , 1995, COLT '95.

[19]  Srinivasan Arunachalam,et al.  Quantum hardness of learning shallow classical circuits , 2019, Electron. Colloquium Comput. Complex..

[20]  Rocco A. Servedio,et al.  Quantum Algorithms for Learning and Testing Juntas , 2007, Quantum Inf. Process..

[21]  Pravesh Kothari,et al.  Constructing Hard Functions Using Learning Algorithms , 2013, 2013 IEEE Conference on Computational Complexity.

[22]  Rocco A. Servedio,et al.  What Circuit Classes Can Be Learned with Non-Trivial Savings? , 2017, ITCS.

[23]  Umesh V. Vazirani,et al.  An Introduction to Computational Learning Theory , 1994 .

[24]  Igor Carboni Oliveira,et al.  An Average-Case Lower Bound against ACC^0 , 2017, Electron. Colloquium Comput. Complex..

[25]  Ryan Williams,et al.  Almost-Everywhere Circuit Lower Bounds from Non-Trivial Derandomization , 2020, 2020 IEEE 61st Annual Symposium on Foundations of Computer Science (FOCS).

[26]  Ryan Williams,et al.  New algorithms and lower bounds for circuits with linear threshold gates , 2014, STOC.

[27]  Shachar Lovett,et al.  Torus polynomials: an algebraic approach to ACC lower bounds , 2018, Electron. Colloquium Comput. Complex..

[28]  Avi Wigderson,et al.  Uniform direct product theorems: simplified, optimized, and derandomized , 2008, SIAM J. Comput..

[29]  Noam Nisan,et al.  Constant depth circuits, Fourier transform, and learnability , 1989, 30th Annual Symposium on Foundations of Computer Science.

[30]  Leonid A. Levin,et al.  A hard-core predicate for all one-way functions , 1989, STOC '89.

[31]  W. Hoeffding Probability Inequalities for sums of Bounded Random Variables , 1963 .

[32]  Igor Carboni Oliveira,et al.  Pseudo-derandomizing learning and approximation , 2018, Electron. Colloquium Comput. Complex..

[33]  Ron Rothblum,et al.  On Exponential-Time Hypotheses, Derandomization, and Circuit Lower Bounds: Extended Abstract , 2020, 2020 IEEE 61st Annual Symposium on Foundations of Computer Science (FOCS).

[34]  Ryan Williams,et al.  Improving exhaustive search implies superpolynomial lower bounds , 2010, STOC '10.

[35]  T. Sanders,et al.  Analysis of Boolean Functions , 2012, ArXiv.

[36]  D. Aharonov A Simple Proof that Toffoli and Hadamard are Quantum Universal , 2003, quant-ph/0301040.

[37]  Lance Fortnow,et al.  Efficient Learning Algorithms Yield Circuit Lower Bounds , 2006, COLT.

[38]  Ryan Williams Nonuniform ACC Circuit Lower Bounds , 2014, JACM.

[39]  Ronald de Wolf,et al.  Guest Column: A Survey of Quantum Learning Theory , 2017, SIGA.

[40]  Ronald de Wolf,et al.  Optimal Quantum Sample Complexity of Learning Algorithms , 2016, CCC.

[41]  Igor Carboni Oliveira,et al.  Conspiracies between Learning Algorithms, Circuit Lower Bounds and Pseudorandomness , 2016, CCC.

[42]  Russell Impagliazzo,et al.  Hard-core distributions for somewhat hard problems , 1995, Proceedings of IEEE 36th Annual Foundations of Computer Science.

[43]  Richard Cleve,et al.  A Quantum Goldreich-Levin Theorem with Cryptographic Applications , 2002, STACS.

[44]  Sanjeev Arora,et al.  Computational Complexity: A Modern Approach , 2009 .

[45]  Iordanis Kerenidis,et al.  Learning with Errors is easy with quantum samples , 2017, Physical Review A.

[46]  Igor Carboni Oliveira,et al.  Randomness and Intractability in Kolmogorov Complexity , 2019, Electron. Colloquium Comput. Complex..

[47]  Ilya Volkovich,et al.  A Guide to Learning Arithmetic Circuits , 2016, COLT.

[48]  Igor Oliveira,et al.  Algorithms versus Circuit Lower Bounds , 2013, Electron. Colloquium Comput. Complex..

[49]  Alan M. Frieze,et al.  Random graphs , 2006, SODA '06.

[50]  Ilya Volkovich On Learning, Lower Bounds and (un)Keeping Promises , 2014, ICALP.

[51]  Alexander A. Razborov,et al.  Natural Proofs , 2007 .

[52]  Ryan O'Donnell,et al.  Efficient quantum tomography II , 2015, STOC.

[53]  Ronald de Wolf,et al.  Two new results about quantum exact learning , 2021, Quantum.

[54]  Cody Murray,et al.  Circuit lower bounds for nondeterministic quasi-polytime: an easy witness lemma for NP and NQP , 2018, Electron. Colloquium Comput. Complex..

[55]  Stephen A. Fenner,et al.  Efficient universal quantum circuits , 2009, Quantum Inf. Comput..