arXiv subjects

Loukas Kavouras

Publications and source records attributed to Loukas Kavouras.

10 recordsLinked to original sources

arXiv2026

Interpretability-by-Design with Accurate Locally Additive Models and Conditional Feature Effects

Generalized additive models (GAMs) offer interpretability through independent univariate feature effects but underfit when interactions are present in data. GA$^2$Ms add selected pairwise interactions which improves accuracy, but sacrifices interpretability and limits model auditing. We propose \emph{Conditionally Additive Local Models} (CALMs), a new model class, that balances the interpretability of GAMs with the accuracy of GA$^2$Ms. CALMs allow multiple univariate shape functions per feature, each active in different regions of the input space. These regions are defined independently for each feature as simple logical conditions (thresholds) on the features it interacts with. As a result, effects remain locally additive while varying across subregions to capture interactions. We further propose a principled distillation-based training pipeline that identifies homogeneous regions with limited interactions and fits interpretable shape functions via region-aware backfitting. Experiments on diverse classification and regression tasks show that CALMs consistently outperform GAMs and achieve accuracy comparable with GA$^2$Ms. Overall, CALMs offer a compelling trade-off between predictive accuracy and interpretability.

cs.LG↗

arXiv2025

Effector: A Python package for regional explanations

Effector is a Python package for interpreting machine learning (ML) models that are trained on tabular data through global and regional feature effects. Global effects, like Partial Dependence Plot (PDP) and Accumulated Local Effects (ALE), are widely used for explaining tabular ML models due to their simplicity -- each feature's average influence on the prediction is summarized by a single 1D plot. However, when features are interacting, global effects can be misleading. Regional effects address this by partitioning the input space into disjoint subregions with minimal interactions within each and computing a separate regional effect per subspace. Regional effects are then visualized by a set of 1D plots per feature. Effector provides efficient implementations of state-of-the-art global and regional feature effects methods under a unified API. The package integrates seamlessly with major ML libraries like scikit-learn and PyTorch. It is designed to be modular and extensible, and comes with comprehensive documentation and tutorials. Effector is an open-source project publicly available on Github at https://github.com/givasile/effector.

cs.LG↗

arXiv2024

GLANCE: Global Actions in a Nutshell for Counterfactual Explainability

The widespread deployment of machine learning systems in critical real-world decision-making applications has highlighted the urgent need for counterfactual explainability methods that operate effectively. Global counterfactual explanations, expressed as actions to offer recourse, aim to provide succinct explanations and insights applicable to large population subgroups. High effectiveness, measured by the fraction of the population that is provided recourse, ensures that the actions benefit as many individuals as possible. Keeping the cost of actions low ensures the proposed recourse actions remain practical and actionable. Limiting the number of actions that provide global counterfactuals is essential to maximizing interpretability. The primary challenge, therefore, is to balance these trade-offs--maximizing effectiveness, minimizing cost, while maintaining a small number of actions. We introduce $\texttt{GLANCE}$, a versatile and adaptive algorithm that employs a novel agglomerative approach, jointly considering both the feature space and the space of counterfactual actions, thereby accounting for the distribution of points in a way that aligns with the model's structure. This design enables the careful balancing of the trade-offs among the three key objectives, with the size objective functioning as a tunable parameter to keep the actions few and easy to interpret. Our extensive experimental evaluation demonstrates that $\texttt{GLANCE}$ consistently shows greater robustness and performance compared to existing methods across various datasets and models.

cs.LG↗

arXiv2024

Fairness in AI: challenges in bridging the gap between algorithms and law

In this paper we examine algorithmic fairness from the perspective of law aiming to identify best practices and strategies for the specification and adoption of fairness definitions and algorithms in real-world systems and use cases. We start by providing a brief introduction of current anti-discrimination law in the European Union and the United States and discussing the concepts of bias and fairness from an legal and ethical viewpoint. We then proceed by presenting a set of algorithmic fairness definitions by example, aiming to communicate their objectives to non-technical audiences. Then, we introduce a set of core criteria that need to be taken into account when selecting a specific fairness definition for real-world use case applications. Finally, we enumerate a set of key considerations and best practices for the design and employment of fairness methods on real-world AI applications

cs.CY↗

arXiv2024

FALE: Fairness-Aware ALE Plots for Auditing Bias in Subgroups

Fairness is steadily becoming a crucial requirement of Machine Learning (ML) systems. A particularly important notion is subgroup fairness, i.e., fairness in subgroups of individuals that are defined by more than one attributes. Identifying bias in subgroups can become both computationally challenging, as well as problematic with respect to comprehensibility and intuitiveness of the finding to end users. In this work we focus on the latter aspects; we propose an explainability method tailored to identifying potential bias in subgroups and visualizing the findings in a user friendly manner to end users. In particular, we extend the ALE plots explainability method, proposing FALE (Fairness aware Accumulated Local Effects) plots, a method for measuring the change in fairness for an affected population corresponding to different values of a feature (attribute). We envision FALE to function as an efficient, user friendly, comprehensible and reliable first-stage tool for identifying subgroups with potential bias issues.

cs.LG↗

arXiv2023

Fairness Aware Counterfactuals for Subgroups

In this work, we present Fairness Aware Counterfactuals for Subgroups (FACTS), a framework for auditing subgroup fairness through counterfactual explanations. We start with revisiting (and generalizing) existing notions and introducing new, more refined notions of subgroup fairness. We aim to (a) formulate different aspects of the difficulty of individuals in certain subgroups to achieve recourse, i.e. receive the desired outcome, either at the micro level, considering members of the subgroup individually, or at the macro level, considering the subgroup as a whole, and (b) introduce notions of subgroup fairness that are robust, if not totally oblivious, to the cost of achieving recourse. We accompany these notions with an efficient, model-agnostic, highly parameterizable, and explainable framework for evaluating subgroup fairness. We demonstrate the advantages, the wide applicability, and the efficiency of our approach through a thorough experimental evaluation of different benchmark datasets.

cs.LG↗

arXiv2022

The Online Min-Sum Set Cover Problem

We consider the online Min-Sum Set Cover (MSSC), a natural and intriguing generalization of the classical list update problem. In Online MSSC, the algorithm maintains a permutation on $n$ elements based on subsets $S_1, S_2, \ldots$ arriving online. The algorithm serves each set $S_t$ upon arrival, using its current permutation $π_{t}$, incurring an access cost equal to the position of the first element of $S_t$ in $π_{t}$. Then, the algorithm may update its permutation to $π_{t+1}$, incurring a moving cost equal to the Kendall tau distance of $π_{t}$ to $π_{t+1}$. The objective is to minimize the total access and moving cost for serving the entire sequence. We consider the $r$-uniform version, where each $S_t$ has cardinality $r$. List update is the special case where $r = 1$. We obtain tight bounds on the competitive ratio of deterministic online algorithms for MSSC against a static adversary, that serves the entire sequence by a single permutation. First, we show a lower bound of $(r+1)(1-\frac{r}{n+1})$ on the competitive ratio. Then, we consider several natural generalizations of successful list update algorithms and show that they fail to achieve any interesting competitive guarantee. On the positive side, we obtain a $O(r)$-competitive deterministic algorithm using ideas from online learning and the multiplicative weight updates (MWU) algorithm. Furthermore, we consider efficient algorithms. We propose a memoryless online algorithm, called Move-All-Equally, which is inspired by the Double Coverage algorithm for the $k$-server problem. We show that its competitive ratio is $Ω(r^2)$ and $2^{O(\sqrt{\log n \cdot \log r})}$, and conjecture that it is $f(r)$-competitive. We also compare Move-All-Equally against the dynamic optimal solution and obtain (almost) tight bounds by showing that it is $Ω(r \sqrt{n})$ and $O(r^{3/2} \sqrt{n})$-competitive.

cs.DS↗

arXiv2022

SCHeMa: Scheduling Scientific Containers on a Cluster of Heterogeneous Machines

In the era of data-driven science, conducting computational experiments that involve analysing large datasets using heterogeneous computational clusters, is part of the everyday routine for many scientists. Moreover, to ensure the credibility of their results, it is very important for these analyses to be easily reproducible by other researchers. Although various technologies, that could facilitate the work of scientists in this direction, have been introduced in the recent years, there is still a lack of open source platforms that combine them to this end. In this work, we describe and demonstrate SCHeMa, an open-source platform that facilitates the execution and reproducibility of computational analysis on heterogeneous clusters, leveraging containerization, experiment packaging, workflow management, and machine learning technologies.

cs.DC↗

arXiv2019

Reallocating Multiple Facilities on the Line

We study the multistage $K$-facility reallocation problem on the real line, where we maintain $K$ facility locations over $T$ stages, based on the stage-dependent locations of $n$ agents. Each agent is connected to the nearest facility at each stage, and the facilities may move from one stage to another, to accommodate different agent locations. The objective is to minimize the connection cost of the agents plus the total moving cost of the facilities, over all stages. $K$-facility reallocation was introduced by de Keijzer and Wojtczak, where they mostly focused on the special case of a single facility. Using an LP-based approach, we present a polynomial time algorithm that computes the optimal solution for any number of facilities. We also consider online $K$-facility reallocation, where the algorithm becomes aware of agent locations in a stage-by-stage fashion. By exploiting an interesting connection to the classical $K$-server problem, we present a constant-competitive algorithm for $K = 2$ facilities.

cs.DS↗

arXiv2017

High-dimensional approximate $r$-nets

The construction of $r$-nets offers a powerful tool in computational and metric geometry. We focus on high-dimensional spaces and present a new randomized algorithm which efficiently computes approximate $r$-nets with respect to Euclidean distance. For any fixed $ε>0$, the approximation factor is $1+ε$ and the complexity is polynomial in the dimension and subquadratic in the number of points. The algorithm succeeds with high probability. More specifically, the best previously known LSH-based construction of Eppstein et al.\ \cite{EHS15} is improved in terms of complexity by reducing the dependence on $ε$, provided that $ε$ is sufficiently small. Our method does not require LSH but, instead, follows Valiant's \cite{Val15} approach in designing a sequence of reductions of our problem to other problems in different spaces, under Euclidean distance or inner product, for which $r$-nets are computed efficiently and the error can be controlled. Our result immediately implies efficient solutions to a number of geometric problems in high dimension, such as finding the $(1+ε)$-approximate $k$th nearest neighbor distance in time subquadratic in the size of the input.

cs.CG↗