Alexandre Quemy

Szczepan Konior, Alexandre Quemy, Przemysław Klocek et al.

Understanding the geometric structure of pre-trained language model embeddings matters for interpretability and safety. We ask whether sentence-level classification signal lives in the Riemannian geometry of contextual token embeddings, and probe it by extracting per-token pullback metrics from a learned encoder's analytical Jacobian and aggregating them with the Fréchet mean on the symmetric positive definite (SPD) manifold; we call this procedure Riemannian Mean Pooling (RMP). Across three datasets with non-trivial linguistic structure (CoLA, CREAK, RTE), RMP outperforms Euclidean mean pooling, while on FEVER-Symmetric, a benchmark constructed to remove annotation-driven lexical artifacts, the method correctly stays at chance. Ablations show that a randomly initialised encoder combined with Fréchet aggregation already beats Euclidean pooling on two of the three signal-bearing datasets, localising the source of the gain to the geometric aggregation rather than to learned manifold structure; the trained encoder contributes additional signal specifically on CREAK, the most knowledge-heavy of the three signal-bearing datasets.

Alexandre Quemy, Marc Schoenauer, Johann Dreo

Multi-objective AI planning suffers from a lack of benchmarks exhibiting known Pareto Fronts. In this work, we propose a tunable benchmark generator, together with a dedicated solver that provably computes the true Pareto front of the resulting instances. First, we prove a proposition allowing us to characterize the optimal plans for a constrained version of the problem, and then show how to reduce the general problem to the constrained one. Second, we provide a constructive way to find all the Pareto-optimal plans and discuss the complexity of the algorithm. We provide an implementation that allows the solver to handle realistic instances in a reasonable time. Finally, as a practical demonstration, we used this solver to find all Pareto-optimal plans between the two largest airports in the world, considering the routes between the 50 largest airports, spherical distances between airports and a made-up risk.

Alexandre Quemy

We introduce Intrinsic Green's Learning (IGL), a framework that models a target function on a manifold as the solution to a linear PDE whose source term is learned from data. Rather than approximating the target directly, IGL learns a source and integrates it against a Green's kernel. An encoder discovers a low-dimensional coordinate chart on the manifold where both the source and the kernel decompose as low-rank tensors, collapsing a high-dimensional integral into independent one-dimensional integrals with cost linear in the intrinsic dimension. A two-stage algorithm separates coordinate discovery from source fitting, a near-convex linear solve, preventing the dimensional collapse of joint training. Learnable gates on each coordinate automatically discover the intrinsic dimension of the manifold. We validate IGL on synthetic manifolds and on MNIST, where it simultaneously achieves near-optimal classification and automatic recovery of the intrinsic dimension.

Alexandre Quemy

This paper presents thirteen datasets for binary, multiclass and multilabel classification based on the European Court of Human Rights judgments since its creation. The interest of such datasets is explained through the prism of the researcher, the data scientist, the citizen and the legal practitioner. Contrarily to many datasets, the creation process, from the collection of raw data to the feature transformation, is provided under the form of a collection of fully automated and open-source scripts. It ensures reproducibility and a high level of confidence in the processed data, which is some of the most important issues in data governance nowadays. A first experimental campaign is performed to study some predictability properties and to establish baseline results on popular machine learning algorithms. The results are consistently good across the binary datasets with an accuracy comprised between 75.86% and 98.32% for an average accuracy of 96.45%.

Alexandre Quemy

Machines learning techniques plays a preponderant role in dealing with massive amount of data and are employed in almost every possible domain. Building a high quality machine learning model to be deployed in production is a challenging task, from both, the subject matter experts and the machine learning practitioners. For a broader adoption and scalability of machine learning systems, the construction and configuration of machine learning workflow need to gain in automation. In the last few years, several techniques have been developed in this direction, known as autoML. In this paper, we present a two-stage optimization process to build data pipelines and configure machine learning algorithms. First, we study the impact of data pipelines compared to algorithm configuration in order to show the importance of data preprocessing over hyperparameter tuning. The second part presents policies to efficiently allocate search time between data pipeline construction and algorithm configuration. Those policies are agnostic from the metaoptimizer. Last, we present a metric to determine if a data pipeline is specific or independent from the algorithm, enabling fine-grain pipeline pruning and meta-learning for the coldstart problem.

Alexandre Quemy

Binary classification is one of the most common problem in machine learning. It consists in predicting whether a given element belongs to a particular class. In this paper, a new algorithm for binary classification is proposed using a hypergraph representation. The method is agnostic to data representation, can work with multiple data sources or in non-metric spaces, and accommodates with missing values. As a result, it drastically reduces the need for data preprocessing or feature engineering. Each element to be classified is partitioned according to its interactions with the training set. For each class, a seminorm over the training set partition is learnt to represent the distribution of evidence supporting this class. Empirical validation demonstrates its high potential on a wide range of well-known datasets and the results are compared to the state-of-the-art. The time complexity is given and empirically validated. Its robustness with regard to hyperparameter sensitivity is studied and compared to standard classification methods. Finally, the limitation of the model space is discussed, and some potential solutions proposed.