arXiv subjects

Nathaniel Harms

Publications and source records attributed to Nathaniel Harms.

22 records · Page 2Linked to original sources

arXiv2021

Downsampling for Testing and Learning in Product Distributions

We study distribution-free property testing and learning problems where the unknown probability distribution is a product distribution over $\mathbb{R}^d$. For many important classes of functions, such as intersections of halfspaces, polynomial threshold functions, convex sets, and $k$-alternating functions, the known algorithms either have complexity that depends on the support size of the distribution, or are proven to work only for specific examples of product distributions. We introduce a general method, which we call downsampling, that resolves these issues. Downsampling uses a notion of "rectilinear isoperimetry" for product distributions, which further strengthens the connection between isoperimetry, testing, and learning. Using this technique, we attain new efficient distribution-free algorithms under product distributions on $\mathbb{R}^d$: 1. A simpler proof for non-adaptive, one-sided monotonicity testing of functions $[n]^d \to \{0,1\}$, and improved sample complexity for testing monotonicity over unknown product distributions, from $O(d^7)$ [Black, Chakrabarty, & Seshadhri, SODA 2020] to $\widetilde O(d^3)$. 2. Polynomial-time agnostic learning algorithms for functions of a constant number of halfspaces, and constant-degree polynomial threshold functions. 3. An $\exp(O(d \log(dk)))$-time agnostic learning algorithm, and an $\exp(O(d \log(dk)))$-sample tolerant tester, for functions of $k$ convex sets; and a $2^{\widetilde O(d)}$ sample-based one-sided tester for convex sets. 4. An $\exp(\widetilde O(k \sqrt d))$-time agnostic learning algorithm for $k$-alternating functions, and a sample-based tolerant tester with the same complexity.

cs.DS↗

arXiv2020

VC Dimension and Distribution-Free Sample-Based Testing

We consider the problem of determining which classes of functions can be tested more efficiently than they can be learned, in the distribution-free sample-based model that corresponds to the standard PAC learning setting. Our main result shows that while VC dimension by itself does not always provide tight bounds on the number of samples required to test a class of functions in this model, it can be combined with a closely-related variant that we call "lower VC" (or LVC) dimension to obtain strong lower bounds on this sample complexity. We use this result to obtain strong and in many cases nearly optimal lower bounds on the sample complexity for testing unions of intervals, halfspaces, intersections of halfspaces, polynomial threshold functions, and decision trees. Conversely, we show that two natural classes of functions, juntas and monotone functions, can be tested with a number of samples that is polynomially smaller than the number of samples required for PAC learning. Finally, we also use the connection between VC dimension and property testing to establish new lower bounds for testing radius clusterability and testing feasibility of linear constraint systems.

cs.LG↗

arXiv2019

Universal Communication, Universal Graphs, and Graph Labeling

We introduce a communication model called universal SMP, in which Alice and Bob receive a function $f$ belonging to a family $\mathcal{F}$, and inputs $x$ and $y$. Alice and Bob use shared randomness to send a message to a third party who cannot see $f, x, y$, or the shared randomness, and must decide $f(x,y)$. Our main application of universal SMP is to relate communication complexity to graph labeling, where the goal is to give a short label to each vertex in a graph, so that adjacency or other functions of two vertices $x$ and $y$ can be determined from the labels $\ell(x),\ell(y)$. We give a universal SMP protocol using $O(k^2)$ bits of communication for deciding whether two vertices have distance at most $k$ on distributive lattices (generalizing the $k$-Hamming Distance problem in communication complexity), and explain how this implies an $O(k^2\log n)$ labeling scheme for determining $\mathrm{dist}(x,y) \leq k$ on distributive lattices with size $n$; in contrast, we show that a universal SMP protocol for determining $\mathrm{dist}(x,y) \leq 2$ in modular lattices (a superset of distributive lattices) has super-constant $Ω(n^{1/4})$ communication cost. On the other hand, we demonstrate that many graph families known to have efficient adjacency labeling schemes, such as trees, low-arboricity graphs, and planar graphs, admit constant-cost communication protocols for adjacency. Trees also have an $O(k)$ protocol for deciding $\mathrm{dist}(x,y) \leq k$ and planar graphs have an $O(1)$ protocol for $\mathrm{dist}(x,y) \leq 2$, which implies a new $O(\log n)$ labeling scheme for the same problem on planar graphs.

cs.CC↗

arXiv2018

Testing Halfspaces over Rotation-Invariant Distributions

We present an algorithm for testing halfspaces over arbitrary, unknown rotation-invariant distributions. Using $\tilde O(\sqrt{n}ε^{-7})$ random examples of an unknown function $f$, the algorithm determines with high probability whether $f$ is of the form $f(x) = sign(\sum_i w_ix_i-t)$ or is $ε$-far from all such functions. This sample size is significantly smaller than the well-known requirement of $Ω(n)$ samples for learning halfspaces, and known lower bounds imply that our sample size is optimal (in its dependence on $n$) up to logarithmic factors. The algorithm is distribution-free in the sense that it requires no knowledge of the distribution aside from the promise of rotation invariance. To prove the correctness of this algorithm we present a theorem relating the distance between a function and a halfspace to the distance between their centers of mass, that applies to arbitrary distributions.

cs.DS↗