arXiv subjects

Klara Pakhomenko

Publications and source records attributed to Klara Pakhomenko.

3 recordsLinked to original sources

arXiv2026

The Boolean Power of ReLU

We prove that, on finite simple undirected graphs equipped with a single Boolean node feature, the Boolean queries expressible in $Σ$-MPLang, for any collection $Σ$ of eventually constant activation functions and with arbitrary real coefficients, form a strict subclass of the Boolean queries expressible in ReLU-MPLang. We thereby settle a recently posed open problem: whether ReLU-MPLang is more powerful than trReLU-MPLang when it comes to Boolean queries. In particular, this implies that ReLU-GNNs are strictly more expressive than {TrReLU,id}-GNNs with respect to Boolean queries on Boolean-featured graphs.

cs.LG↗

arXiv2026

A Logical View of GNN-Style Computation and the Role of Activation Functions

We study the numerical and Boolean expressiveness of MPLang, a declarative language that captures the computation of graph neural networks (GNNs) through linear message passing and activation functions. We begin with A-MPLang, the fragment without activation functions, and give a characterization of its expressive power in terms of walk-summed features. For bounded activation functions, we show that (under mild conditions) all eventually constant activations yield the same expressive power - numerical and Boolean - and that it subsumes previously established logics for GNNs with eventually constant activation functions but without linear layers. Finally, we prove the first expressive separation between unbounded and bounded activations in the presence of linear layers: MPLang with ReLU is strictly more powerful for numerical queries than MPLang with eventually constant activation functions, e.g., truncated ReLU. This hinges on subtle interactions between linear aggregation and eventually constant non-linearities, and it establishes that GNNs using ReLU are more expressive than those restricted to eventually constant activations and linear layers.

cs.LG↗

arXiv2026

Weisfeiler-Leman on graphs of small twin-width

Twin-width is a graph parameter introduced in the context of first-order model checking, and has since become a central parameter in algorithmic graph theory. While many algorithmic problems become easier on arbitrary classes of bounded twin-width, graph isomorphism on graphs of twin-width 4 and above is as hard as the general isomorphism problem. For each positive number $k$, the $k$-dimensional Weisfeiler-Leman algorithm is an iterative color refinement algorithm that encodes structural similarities and serves as a fundamental tool for distinguishing non-isomorphic graphs. We show that the graph isomorphism problem for graphs of twin-width 1 can be solved by the purely combinatorial 3-dimensional Weisfeiler-Leman algorithm, while there is no fixed $k$ such that the $k$-dimensional Weisfeiler-Leman algorithm solves the graph isomorphism problem for graphs of twin-width 4. Moreover, we prove the conjecture of Bergougnoux, Gajarský, Guspiel, Hlinený, Pokrývka, and Sokolowski that stable graphs of twin-width 2 have bounded rank-width. This in particular implies that isomorphism of these graphs can be decided by a fixed dimension of the Weisfeiler-Leman algorithm.

math.CO↗