Machine learning models now drive many critical decisions, making explanations of their reasoning essential. Recent work analyzes the complexity of exact explanations in transparent models, but these explanations are often too large for practical use. This has motivated research into probabilistic alternatives. We study probabilistic extensions that allow controlled uncertainty while maintaining rigorous foundations. We analyze three basic model types: decision trees, decision lists, and decision sets. We introduce algorithms for computing both local and global probabilistic explanations for these models. Our main result shows that computing minimum-size probabilistic explanations is fixed-parameter tractable when parameterized by structural properties---specifically, the number of terms for decision lists and decision sets and the minimum of the number of positive and the number of negative leaves.

Determining the validity of a quantified Boolean formula (QBF) is a PSPACE-complete problem with rich expressive power. Despite interest in efficient solvers, there is, compared to problems in NP, a lack of positive theoretical results, and in the parameterized complexity setting one often has to restrict the quantifier prefix (e.g., bounding alternations) to obtain fixed parameter tractability (FPT). We propose a new parameter: the number of variables in clauses that has to be removed before reaching a tractable class (a clause covering (CC) backdoor). We are then interested in solving QBF in FPT time given a CC-backdoor of size k. We consider the three classical, tractable cases of QBF as base classes: Horn, 2-CNF, and linear equations. We establish W[1]-hardness for Horn but prove FPT for the others, and prove that in a precise, algebraic sense, we are only missing one important case for a full dichotomy. Our algorithms are non-trivial and depend on propagation, and Gaussian elimination, respectively, and are comparably unexplored for QBF.

We develop a general algorithmic framework that allows us to obtain fixed-parameter tractability for computing smallest symbolic models that represent given data. Our framework applies to all ML model types that admit a certain extension property. By establishing this extension property for decision trees, decision sets, decision lists, and binary decision diagrams, we obtain that minimizing these fundamental model types is fixed-parameter tractable. Our framework even applies to ensembles, which combine individual models by majority decision.

We develop a general algorithmic framework that allows us to obtain fixed-parameter tractability for computing smallest symbolic models that represent given data. Our framework applies to all ML model types that admit a certain extension property. By showing this extension property for decision trees, decision sets, decision lists, and binary decision diagrams, we obtain that minimizing these fundamental model types is fixed-parameter tractable. Our framework even applies to ensembles, which combine individual models by majority decision.

A book embedding of a graph is a drawing that maps vertices onto a line and edges to simple pairwise non-crossing curves drawn into "pages", which are half-planes bounded by that line. Two-page book embeddings, i.e., book embeddings into 2 pages, are of special importance as they are both NP-hard to compute and have specific applications. We obtain a 2^𝒪(√n) algorithm for computing a book embedding of an n-vertex graph on two pages - a result which is asymptotically tight under the Exponential Time Hypothesis. As a key tool in our approach, we obtain a single-exponential fixed-parameter algorithm for the same problem when parameterized by the treewidth of the input graph. We conclude by establishing the fixed-parameter tractability of computing minimum-page book embeddings when parameterized by the feedback edge number, settling an open question arising from previous work on the problem.

This paper presents a comprehensive theoretical investigation into the parameterized complexity of explanation problems in various machine learning (ML) models. Contrary to the prevalent black-box perception, our study focuses on models with transparent internal mechanisms. We address two principal types of explanation problems: abductive and contrastive, both in their local and global variants. Our analysis encompasses diverse ML models, including Decision Trees, Decision Sets, Decision Lists, Ordered Binary Decision Diagrams, Random Forests, and Boolean Circuits, and ensembles thereof, each offering unique explanatory challenges. This research fills a significant gap in explainable AI (XAI) by providing a foundational understanding of the complexities of generating explanations for these models. This work provides insights vital for further research in the domain of XAI, contributing to the broader discourse on the necessity of transparency and accountability in AI systems.

Abstract We present a flexible and efficient toolchain to symbolically solve (standard) Rabin games, fair-adversarial Rabin games, and "Image missing" -player Rabin games. To our best knowledge, our tools are the first ones to be able to solve these problems. Furthermore, using these flexible game solvers as a back-end, we implemented a tool for computing correct-by-construction controllers for stochastic dynamical systems under LTL specifications. Our implementations use the recent theoretical result that all of these games can be solved using the same symbolic fixpoint algorithm but utilizing different, domain specific calculations of the involved predecessor operators. The main feature of our toolchain is the utilization of two programming abstractions: one to separate the symbolic fixpoint computations from the predecessor calculations, and another one to allow the integration of different BDD libraries as back-ends. In particular, we employ a multi-threaded execution of the fixpoint algorithm by using the multi-threaded BDD library Sylvan, which leads to enormous computational savings.

We present a flexible and efficient toolchain to symbolically solve (standard) Rabin games, fair-adversarial Rabin games, and 21/2-player Rabin games. To our best knowledge, our tools are the first ones to be able to solve these problems. Furthermore, using the optimized game solvers as back-end, we implement a tool for computing correct-by-construction controllers for stochastic dynamical systems with LTL specifications. An important feature of our toolchain is the flexibility created through two programming abstractions: one separates the symbolic fixpoint computations from the predecessor calculations, and the other one allows effortless switching between different BDD libraries. We empirically compare the benefits of using the CUDD and Sylvan BDD libraries, and report substantial computational savings of our tool compared to the state-of-the-art.

We revisit classic string problems considered in the area of parameterized complexity, and study them through the lens of dynamic data structures. That is, instead of asking for a static algorithm that solves the given instance efficiently, our goal is to design a data structure that efficiently maintains a solution, or reports a lack thereof, upon updates in the instance. We first consider the Closest String problem, for which we design randomized dynamic data structures with amortized update times $d^{\mathcal{O}(d)}$ and $|Σ|^{\mathcal{O}(d)}$, respectively, where $Σ$ is the alphabet and $d$ is the assumed bound on the maximum distance. These are obtained by combining known static approaches to Closest String with color-coding. Next, we note that from a result of Frandsen et al.~[J. ACM'97] one can easily infer a meta-theorem that provides dynamic data structures for parameterized string problems with worst-case update time of the form $\mathcal{O}(\log \log n)$, where $k$ is the parameter in question and $n$ is the length of the string. We showcase the utility of this meta-theorem by giving such data structures for problems Disjoint Factors and Edit Distance. We also give explicit data structures for these problems, with worst-case update times $\mathcal{O}(k2^{k}\log \log n)$ and $\mathcal{O}(k^2\log \log n)$, respectively. Finally, we discuss how a lower bound methodology introduced by Amarilli et al.~[ICALP'21] can be used to show that obtaining update time $\mathcal{O}(f(k))$ for Disjoint Factors and Edit Distance is unlikely already for a constant value of the parameter $k$.