Abstract Explainable recommendation systems enhance user trust and satisfaction by revealing the reasoning behind personalized recommendations. Approaching this as a post-hoc explanation-ranking problem over a fixed pool of candidate explanations, we propose Contrastive Learning for Explanation Ranking (CLER), a model that learns user, item, and explanation representations with a Normalized Temperature-scaled Binary Cross-Entropy (NT-BXent) loss. This function specifically applies a per-row reweighting strategy, preventing the vast number of negative examples from dominating the objective. We evaluate CLER on the Amazon, TripAdvisor, and Yelp datasets from the EXTRA benchmark. Across traditional ranking metrics, CLER achieves the strongest results among the compared baselines.

[Abstract]: In recent years, the Natural Language Processing (NLP) field has experienced a revolution, where numerous models – based on the Transformer architecture – have emerged to process the ever-growing volume of online text-generated data. This architecture has been the basis for the rise of Large Language Models (LLMs). Enabling their application to many diverse tasks in which they excel with just a fine-tuning process that comes right after a vast pre-training phase. However, their sustainability can often be overlooked, especially regarding computational and environmental costs. Our research aims to compare various BERT derivatives in the context of a dyadic data task while also drawing attention to the growing need for sustainable AI solutions. To this end, we utilize a selection of transformer models in an explainable recommendation setting, modeled as a multi-label classification task originating from a social network context, where users, restaurants, and reviews interact.

Dyadic regression models, which output real-valued predictions for pairs of entities, are fundamental in many domains [e.g., obtaining user-product ratings in recommender systems (RSs)] and promising and under exploration in others (e.g., tuning patient-drug dosages in precision pharmacology). In this work, we prove that nonuniform observed value distributions of individual entities lead to severe biases in state-of-the-art models, skewing predictions toward the average of observed past values for the entity and providing worse-than-random predictive power in eccentric yet crucial cases; we name this phenomenon eccentricity bias. We show that global error metrics like root-mean-squared error (RMSE) are insufficient to capture this bias, and we introduce eccentricity area under the curve (EAUC) as a novel metric that can quantify it in all studied domains and models. We prove the intuitive interpretation of EAUC by experimenting with naive post-training bias corrections and theorize other options to use EAUC to guide the construction of fair models. This work contributes a bias-aware evaluation of dyadic regression to prevent unfairness in critical real-world applications of such systems.

Green artificial intelligence (AI) is more environmentally friendly and inclusive than conventional AI, as it not only produces accurate results without increasing the computational cost but also ensures that any researcher with a laptop can perform high-quality research without the need for costly cloud servers. This paper discusses green AI as a pivotal approach to enhancing the environmental sustainability of AI systems. Described are AI solutions for eco-friendly practices in other fields (green-by AI), strategies for designing energy-efficient machine learning (ML) algorithms and models (green-in AI), and tools for accurately measuring and optimizing energy consumption. Also examined are the role of regulations in promoting green AI and future directions for sustainable ML. Underscored is the importance of aligning AI practices with environmental considerations, fostering a more eco-conscious and energy-efficient future for AI systems.

Green machine learning refers to research that is more environmentally friendly and inclusive, not only by producing novel results without increasing the computational cost, but also by ensuring that any researcher with a laptop has the opportunity to perform high-quality research without the need to use expensive cloud servers.Efficient machine learning approaches (especially deep learning) are starting to receive some attention in the research community.This tutorial is concerned with the development of machine learning algorithms that optimize efficiency rather than only accuracy.We provide an overview of this recent field, together with a review of the novel contributions to the ESANN 2023 special session on Green Machine Learning. * This work

Classic embedded feature selection algorithms are often divided in two large groups: tree-based algorithms and LASSO variants. Both approaches are focused in different aspects: while the tree-based algorithms provide a clear explanation about which variables are being used to trigger a certain output, LASSO-like approaches sacrifice a detailed explanation in favor of increasing its accuracy. In this paper, we present a novel embedded feature selection algorithm, called End-to-End Feature Selection (E2E-FS), that aims to provide both accuracy and explainability in a clever way. Despite having non-convex regularization terms, our algorithm, similar to the LASSO approach, is solved with gradient descent techniques, introducing some restrictions that force the model to specifically select a maximum number of features that are going to be used subsequently by the classifier. Although these are hard restrictions, the experimental results obtained show that this algorithm can be used with any learning model that is trained using a gradient descent algorithm.

In this study, we analyze the capability of several state of the art machine learning methods to predict whether patients diagnosed with CoVid-19 (CoronaVirus disease 2019) will need different levels of hospital care assistance (regular hospital admission or intensive care unit admission), during the course of their illness, using only demographic and clinical data. For this research, a data set of 10,454 patients from 14 hospitals in Galicia (Spain) was used. Each patient is characterized by 833 variables, two of which are age and gender and the other are records of diseases or conditions in their medical history. In addition, for each patient, his/her history of hospital or intensive care unit (ICU) admissions due to CoVid-19 is available. This clinical history will serve to label each patient and thus being able to assess the predictions of the model. Our aim is to identify which model delivers the best accuracies for both hospital and ICU admissions only using demographic variables and some structured clinical data, as well as identifying which of those are more relevant in both cases. The results obtained in the experimental study show that the best models are those based on oversampling as a preprocessing phase to balance the distribution of classes. Using these models and all the available features, we achieved an area under the curve (AUC) of 76.1% and 80.4% for predicting the need of hospital and ICU admissions, respectively. Furthermore, feature selection and oversampling techniques were applied and it has been experimentally verified that the relevant variables for the classification are age and gender, since only using these two features the performance of the models is not degraded for the two mentioned prediction problems.

This paper presents a unified propagation method for dealing with both the classic Eikonal equation, where the motion direction does not affect the propagation, and the more general static Hamilton-Jacobi equations, where it does. While classic Fast Marching Method (FMM) techniques achieve the solution to the Eikonal equation with a O(M log M) (or O(M) assuming some modifications), solving the more general static Hamilton-Jacobi equation requires a higher complexity. The proposed framework maintains the O(M log M) complexity for both problems, while achieving higher accuracy than available state-of-the-art. The key idea behind the proposed method is the creation of 'mini wave-fronts', where the solution is interpolated to minimize the discretization error. Experimental results show how our algorithm can outperform the state-of-the-art both in precision and computational cost.

With the rise of the so-called Adversarial Attacks, there is an increased concern on model security. In this paper we present two different contributions: novel measures of robustness (based on adversarial attacks) and a novel adversarial attack. The key idea behind these metrics is to obtain a measure that could compare different architectures, with independence of how the input is preprocessed (robustness against different input sizes and value ranges). To do so, a novel adversarial attack is presented, performing a delayed elastic-net adversarial attack (constraints are only used whenever a successful adversarial attack is obtained). Experimental results show that our approach obtains state-of-the-art adversarial samples, in terms of minimal perturbation distance. Finally, a benchmark of ImageNet pretrained models is used to conduct experiments aiming to shed some light about which model should be selected whenever security is a role factor.

Abstract Many aspects of our lives are associated with places and the activities we perform on a daily basis. Most of them are recurrent and demand displacement of the individual between regular places like going to work, school or other important personal locations. To accomplish these recurrent daily activities, people tend to follow regular paths with similar temporal and spatial characteristics, especially because humans are frequently looking for uniformity to support their decisions and make their actions easier or even automatic. In this work, we propose a method for discovering common pathways across users' habits from human mobility data. By using a density‐based clustering algorithm, we identify the most preferable locations the users visit, we apply a Gaussian mixture model over these places to automatically separate among all traces, the trajectories that follow patterns in order to discover the representations of individual's habits. By using the longest common sub‐sequence algorithm, we search for the trajectories that are more similar over the set of users' habits trips by considering the distance that pairs of users or habits share on the same path. The proposed method is evaluated over two real‐world GPS datasets and the results show that the approach is able to detect the most important places in a user's life, detect the routine activities and identify common routes between users that have similar habits paving the way for research techniques in carpooling, recommendation and prediction systems.

This paper presents a new wavefront propagation method for dealing with the classic Eikonal equation. While classic Dijkstra-like graph-based techniques achieve the solution in O(M log M), they do not approximate the unique physically relevant solution very well. Fast Marching Methods (FMM) were created to efficiently solve the continuous problem. The proposed approximation tries to maintain the complexity, in order to make the algorithm useful in a wide range of contexts. The key idea behind our method is the creation of 'mini wave-fronts', which are combined to propagate the solution. Experimental results show the improvement in the accuracy with respect to the state of the art, while the average computational speed is maintained in O(M log M), similar to the FMM techniques.

This paper presents a novel methodology for modelling pedestrian trajectories over a scene, based in the hypothesis that, when people try to reach a destination, they use the path that takes less time, taking into account environmental information like the type of terrain or what other people did before. Thus, a minimal path approach can be used to model human trajectory behaviour. We develop a modified Fast Marching Method that allows us to include both velocity and orientation in the Front Propagation Approach, without increasing its computational complexity. Combining all the information, we create a time surface that shows the time a target need to reach any given position in the scene. We also create different metrics in order to compare the time surface against the real behaviour. Experimental results over a public dataset prove the initial hypothesis' correctness.

Crossovers and bifurcations are interest points of the retinal vascular tree useful to diagnose diseases. Specifically, detecting these interest points and identifying which of them are crossings will give us the opportunity to search for arteriovenous nicking, this is, an alteration of the vessel tree where an artery is crossed by a vein and the former compresses the later. These formations are a clear indicative of hypertension, among other medical problems. There are several studies that have attempted to define an accurate and reliable method to detect and classify these relevant points. In this article, we propose a new method to identify crossovers. Our approach is based on segmenting the vascular tree and analyzing the surrounding area of each interest point. The minimal path between vessel points in this area is computed in order to identify the connected vessel segments and, as a result, to distinguish between bifurcations and crossovers. Our method was tested using retinographies from public databases DRIVE and VICAVR, obtaining an accuracy of 90%.