The Taub Faculty of Computer Science Events and Talks

We develop a framework for proving lower bounds on computational problems over distributions, including optimization and unsupervised learning. Our framework is based on defining a restricted class of algorithms, called statistical algorithms, that instead of accessing samples from the input distribution can only obtain an estimate of the expectation of any given function on a sample drawn randomly from the input distribution.

Our definition captures many natural algorithms used in theory and practice, e.g. moments-based methods, local search, MCMC and simulated annealing. Our techniques are inspired by (and generalize) the statistical query model in learning theory, which captures the complexity of PAC learning using essentially all known learning methods (Kearns, JACM 1998).

For specific well-known problems over distributions, we give lower bounds on the complexity of any statistical algorithm. These include an exponential lower bounds for moment maximization in R^n, and a nearly optimal lower bound for detecting planted clique distributions when the planted clique has size $O(n^{1/2-\delta})$ for any constant $\delta > 0$. Variants of the latter problem have been assumed to be hard to prove hardness for other problems and for cryptographic applications. Our lower bounds provide the first concrete evidence supporting these assumptions.

Joint work with Elena Grigorescu, Lev Reyzin, Santosh Vempala and Ying Xiao.