arXiv subjects
Publications and source records attributed to Maryanthe Malliaris.
We show that certain families of sets in $\mathbb{R}^2$ (or $\mathbb{R}^n$) which are neither definable nor have bounded VC-dimension are nonetheless uniformly approximately definable in the real field, an o-minimal structure.
In this brief note, we first give a counterexample to a theorem in Chernikov and Towsner, arXiv:2510.02420(1). In arXiv:2510.02420(2), the theorem has changed but as we explain the proof has a mistake. The change in the statement, due to changes in the underlying definition, affects the paper's claims. Since that theorem had been relevant to connecting the work of their paper to Coregliano-Malliaris high-arity PAC learning, a connection which now disappears, we also explain why their definitions miss crucial aspects that our work was designed to grapple with.
We develop a new high-dimensional statistical learning model which can take advantage of structured correlation in data even in the presence of randomness. We completely characterize learnability in this model in terms of VCN${}_{k,k}$-dimension (essentially $k$-dependence from Shelah's classification theory). This model suggests a theoretical explanation for the success of certain algorithms in the 2006~Netflix Prize competition.
We give a model-theoretic perspective on regular ultrafilter construction in the twentieth and twenty-first century (so far), and explain the "canonical Boolean algebra" recently developed by Malliaris and Shelah.
Any Littlestone class, or stable graph, has finite sets which function as ``virtual elements'': these can be seen from the learning side as representing hypotheses which are expressible as weighted majority opinions of hypotheses in the class, and from the model-theoretic side as an approximate finitary version of realizing types. We introduce and study the epsilon-saturation of a Littlestone class, or stable graph, which is essentially the closure of the class under inductively adding all such virtual elements. We characterize this closure and prove that under reasonable choices of parameters, it remains Littlestone (or stable), though not always of the same Littlestone dimension. This highlights some surprising phenomena having to do with regimes of epsilon and the relation between Littlestone/stability and VC dimension.
We use algorithmic methods from online learning to explore some important objects at the intersection of model theory and combinatorics, and find natural ways that algorithmic methods can detect and explain (and improve our understanding of) stable structure in the sense of model theory. The main theorem deals with existence of $ε$-excellent sets (which are key to the Stable Regularity Lemma, a theorem characterizing the appearance of irregular pairs in Szemerédi's celebrated Regularity Lemma). We prove that $ε$-excellent sets exist for any $ε< \frac{1}{2}$ in $k$-edge stable graphs in the sense of model theory (equivalently, Littlestone classes); earlier proofs had given this only for $ε< 1/{2^{2^k}}$ or so. We give two proofs: the first uses regret bounds from online learning, the second uses Boolean closure properties of Littlestone classes and sampling. We also give a version of the dynamic Sauer-Shelah-Perles lemma appropriate to this setting, related to definability of types. We conclude by characterizing stable/Littlestone classes as those supporting a certain abstract notion of majority: the proof shows that the two distinct, natural notions of majority, arising from measure and from dimension, densely often coincide.
This paper is about the surprising interaction of a foundational result from model theory, about stability of theories, with algorithmic stability in learning. First, in response to gaps in existing learning models, we introduce a new statistical learning model, called ``Probably Eventually Correct'' or PEC. We characterize Littlestone (stable) classes in terms of this model. As a corollary, Littlestone classes have frequent short definitions in a natural statistical sense. In order to obtain a characterization of Littlestone classes in terms of frequent definitions, we build an equivalence theorem highlighting what is common to many existing approximation algorithms, and to the new PEC. This is guided by an analogy to definability of types in model theory, but has its own character. Drawing on these theorems and on other recent work, we present a complete algorithmic analogue of Shelah's celebrated Unstable Formula Theorem, with algorithmic properties taking the place of the infinite.
Recently, the authors introduced the theory of high-arity PAC learning, which is well-suited for learning graphs, hypergraphs and relational structures. In the same initial work, the authors proved a high-arity analogue of the Fundamental Theorem of Statistical Learning that almost completely characterizes all notions of high-arity PAC learning in terms of a combinatorial dimension, called the Vapnik--Chervonenkis--Natarajan (VCN${}_k$) $k$-dimension, leaving as an open problem only the characterization of non-partite, non-agnostic high-arity PAC learnability. In this work, we complete this characterization by proving that non-partite non-agnostic high-arity PAC learnability implies a high-arity version of the Haussler packing property, which in turn implies finiteness of VCN${}_k$-dimension. This is done by obtaining direct proofs that classic PAC learnability implies classic Haussler packing property, which in turn implies finite Natarajan dimension and noticing that these direct proofs nicely lift to high-arity.
We develop a theory of high-arity PAC learning, which is statistical learning in the presence of "structured correlation". In this theory, hypotheses are either graphs, hypergraphs or, more generally, structures in finite relational languages, and i.i.d. sampling is replaced by sampling an induced substructure, producing an exchangeable distribution. Our main theorems establish a high-arity (agnostic) version of the fundamental theorem of statistical learning.
Call a hereditary family $\mathcal{F}$ of graphs strongly persistent if there exists a graphon $W$ such that in all subgraphons $W'$ of $W$, $\mathcal{F}$ is precisely the class of finite graphs that have positive density in $W'$. Our first result is a complete characterization of the hereditary families of graphs that are strongly persistent as precisely those that are closed under substitutions. We call graphons with the self-similarity property above weakly random. A hereditary family $\mathcal{F}$ is said to have the weakly random Erdős--Hajnal property (WR) if every graphon that is a limit of graphs in $\mathcal{F}$ has a weakly random subgraphon. Among families of graphs that are closed under substitutions, we completely characterize the families that belong to WR as those with "few" prime graphs. We also extend some of the results above to structures in finite relational languages by using the theory of theons.
What happens when mathematics realizes infinity. When are mathematical definitions actually useful?
The celebrated Erdős-Hajnal Conjecture says that in any proper hereditary class of finite graphs we are guaranteed to have a clique or anti-clique of size $n^c$, which is a much better bound than the logarithmic size that is provided by Ramsey's Theorem in general. On the other hand, in uncountable cardinalities, the model-theoretic property of stability guarantees a uniform set much larger than the bound provided by the Erdős-Rado Theorem in general. Even though the consequences of stability in the finite have been much studied in the literature, the countable setting seems a priori quite different, namely, in the countably infinite the notion of largeness based on cardinality alone does not reveal any structure as Ramsey's Theorem already provides a countably infinite uniform set in general. In this paper, we show that the natural notion of largeness given by upper density reveals that these phenomena meet in the countable: a countable graph has an almost clique or anti-clique of positive upper density if and only if it has a positive upper density almost stable set. Moreover, this result also extends naturally to countable models of a universal theory in a finite relational language. Our methods explore a connection with the notion of convergence in the theory of limits of dense combinatorial objects, introducing and studying a natural approximate version of the Erdős-Hajnal property that allows for a negligible error in the edges (in general, predicates) but requires linear-sized uniform sets in convergent sequences of models (this is much stronger than what stable regularity can provide as the error is required to go to zero). Finally, surprisingly, we completely characterize all hereditary classes of finite graphs that have this approximate Erdős-Hajnal property. The proof highlights both differences and similarities with the original conjecture.
By studying the commuting graphs of conjugacy classes of the sequence of Heisenberg groups $H_{2n+1}(p)$ and their limit $H_\infty(p)$ we find pseudo-random behavior (and the random graph in the limiting case). This makes a nice case study for transfer of information between finite and infinite objects. Some of this behavior transfers to the problem of understanding what makes understanding the character theory of the uni-upper-triangular group (mod p) "wild". Our investigations in this paper may be seen as a meditation on the question: is randomness simple or is it complicated?
We show that every approximately differentially private learning algorithm (possibly improper) for a class $H$ with Littlestone dimension~$d$ requires $Ω\bigl(\log^*(d)\bigr)$ examples. As a corollary it follows that the class of thresholds over $\mathbb{N}$ can not be learned in a private manner; this resolves open question due to [Bun et al., 2015, Feldman and Xiao, 2015]. We leave as an open question whether every class with a finite Littlestone dimension can be learned by an approximately differentially private algorithm.
These notes are based on Appalachian Set Theory lectures given by M. Malliaris on November 5, 2016 with D. Casey as the official scribe. The aim of the lectures was to present the setup and some key arguments of "Cofinality spectrum problems in model theory, set theory and general topology" by Malliaris and Shelah. This provides a sketch of the proof that $\mathfrak{p} = \mathfrak{t}$ and that $SOP_2$ theories are maximal in Keisler's Order.
We prove a regularity lemma with respect to arbitrary Keisler measures mu on V, nu on W where the bipartite graph (V,W,R) is definable in a saturated structure M and the formula R(x,y) is stable. The proof is rather quick and uses local stability theory. The special case where (V,W,R) is pseudofinite, mu, nu are the counting measures and M is suitably chosen (for example a nonstandard model of set theory), yields the stable regularity theorem of Malliaris-Shelah (Transactions AMS, 366, 2014, 1551-1585), though without explicit bounds or equitability.