Denis Parra

Pablo Messina, René Vidal, Denis Parra et al.

Advancing representation learning in specialized fields like medicine remains challenging due to the scarcity of expert annotations for text and images. To tackle this issue, we present a novel two-stage framework designed to extract high-quality factual statements from free-text radiology reports in order to improve the representations of text encoders and, consequently, their performance on various downstream tasks. In the first stage, we propose a \textit{Fact Extractor} that leverages large language models (LLMs) to identify factual statements from well-curated domain-specific datasets. In the second stage, we introduce a \textit{Fact Encoder} (CXRFE) based on a BERT model fine-tuned with objective functions designed to improve its representations using the extracted factual data. Our framework also includes a new embedding-based metric (CXRFEScore) for evaluating chest X-ray text generation systems, leveraging both stages of our approach. Extensive evaluations show that our fact extractor and encoder outperform current state-of-the-art methods in tasks such as sentence ranking, natural language inference, and label extraction from radiology reports. Additionally, our metric proves to be more robust and effective than existing metrics commonly used in the radiology report generation literature. The code of this project is available at \url{https://github.com/PabloMessina/CXR-Fact-Encoder}.

Ivania Donoso-Guzmán, Jeroen Ooge, Denis Parra et al.

While research on explainable AI (XAI) is booming and explanation techniques have proven promising in many application domains, standardised human-centred evaluation procedures are still missing. In addition, current evaluation procedures do not assess XAI methods holistically in the sense that they do not treat explanations' effects on humans as a complex user experience. To tackle this challenge, we propose to adapt the User-Centric Evaluation Framework used in recommender systems: we integrate explanation aspects, summarise explanation properties, indicate relations between them, and categorise metrics that measure these properties. With this comprehensive evaluation framework, we hope to contribute to the human-centred standardisation of XAI evaluation.

Rafael Elberg, Denis Parra, Mircea Petrache

Image and multimodal machine learning tasks are very challenging to solve in the case of poorly distributed data. In particular, data availability and privacy restrictions exacerbate these hurdles in the medical domain. The state of the art in image generation quality is held by Latent Diffusion models, making them prime candidates for tackling this problem. However, a few key issues still need to be solved, such as the difficulty in generating data from under-represented classes and a slow inference process. To mitigate these issues, we propose a new method for image augmentation in long-tailed data based on leveraging the rich latent space of pre-trained Stable Diffusion Models. We create a modified separable latent space to mix head and tail class examples. We build this space via Iterated Learning of underlying sparsified embeddings, which we apply to task-specific saliency maps via a K-NN approach. Code is available at https://github.com/SugarFreeManatee/Feature-Space-Augmentation-and-Iterated-Learning

Ivania Donoso-Guzmán, Kristýna Sirka Kacafírková, Maxwell Szymanski et al.

Despite promising developments in Explainable Artificial Intelligence, the practical value of XAI methods remains under-explored and insufficiently validated in real-world settings. Robust and context-aware evaluation is essential, not only to produce understandable explanations but also to ensure their trustworthiness and usability for intended users, but tends to be overlooked because of no clear guidelines on how to design an evaluation with users. This study addresses this gap with two main goals: (1) to develop a framework of well-defined, atomic properties that characterise the user experience of XAI in healthcare; and (2) to provide clear, context-sensitive guidelines for defining evaluation strategies based on system characteristics. We conducted a systematic review of 82 user studies, sourced from five databases, all situated within healthcare settings and focused on evaluating AI-generated explanations. The analysis was guided by a predefined coding scheme informed by an existing evaluation framework, complemented by inductive codes developed iteratively. The review yields three key contributions: (1) a synthesis of current evaluation practices, highlighting a growing focus on human-centred approaches in healthcare XAI; (2) insights into the interrelations among explanation properties; and (3) an updated framework and a set of actionable guidelines to support interdisciplinary teams in designing and implementing effective evaluation strategies for XAI systems tailored to specific application contexts.

Mingquan Lin, Gregory Holste, Song Wang et al.

The CXR-LT series is a community-driven initiative designed to enhance lung disease classification using chest X-rays (CXR). It tackles challenges in open long-tailed lung disease classification and enhances the measurability of state-of-the-art techniques. The first event, CXR-LT 2023, aimed to achieve these goals by providing high-quality benchmark CXR data for model development and conducting comprehensive evaluations to identify ongoing issues impacting lung disease classification performance. Building on the success of CXR-LT 2023, the CXR-LT 2024 expands the dataset to 377,110 chest X-rays (CXRs) and 45 disease labels, including 19 new rare disease findings. It also introduces a new focus on zero-shot learning to address limitations identified in the previous event. Specifically, CXR-LT 2024 features three tasks: (i) long-tailed classification on a large, noisy test set, (ii) long-tailed classification on a manually annotated "gold standard" subset, and (iii) zero-shot generalization to five previously unseen disease findings. This paper provides an overview of CXR-LT 2024, detailing the data curation process and consolidating state-of-the-art solutions, including the use of multimodal models for rare disease detection, advanced generative approaches to handle noisy labels, and zero-shot learning strategies for unseen diseases. Additionally, the expanded dataset enhances disease coverage to better represent real-world clinical settings, offering a valuable resource for future research. By synthesizing the insights and innovations of participating teams, we aim to advance the development of clinically realistic and generalizable diagnostic models for chest radiography.

Gregory Schuit, Denis Parra, Cecilia Besa

Generative image models have achieved remarkable progress in both natural and medical imaging. In the medical context, these techniques offer a potential solution to data scarcity-especially for low-prevalence anomalies that impair the performance of AI-driven diagnostic and segmentation tools. However, questions remain regarding the fidelity and clinical utility of synthetic images, since poor generation quality can undermine model generalizability and trust. In this study, we evaluate the effectiveness of state-of-the-art generative models-Generative Adversarial Networks (GANs) and Diffusion Models (DMs)-for synthesizing chest X-rays conditioned on four abnormalities: Atelectasis (AT), Lung Opacity (LO), Pleural Effusion (PE), and Enlarged Cardiac Silhouette (ECS). Using a benchmark composed of real images from the MIMIC-CXR dataset and synthetic images from both GANs and DMs, we conducted a reader study with three radiologists of varied experience. Participants were asked to distinguish real from synthetic images and assess the consistency between visual features and the target abnormality. Our results show that while DMs generate more visually realistic images overall, GANs can report better accuracy for specific conditions, such as absence of ECS. We further identify visual cues radiologists use to detect synthetic images, offering insights into the perceptual gaps in current models. These findings underscore the complementary strengths of GANs and DMs and point to the need for further refinement to ensure generative models can reliably augment training datasets for AI diagnostic systems.

Rafael Elberg, Felipe del Rio, Mircea Petrache et al.

Compositional generalization--the ability to interpret novel combinations of familiar elements--is a hallmark of human cognition and language. Despite recent advances, deep neural networks still struggle to acquire this property reliably. In this work, we introduce CELEBI (Compressive-Expressive Language Emergence through a discrete Bottleneck and Iterated learning), a novel self-supervised framework for inducing compositionality in learned representations from pre-trained models, through a reconstruction-based communication game between a sender and a receiver. Building on theories of language emergence, we integrate three mechanisms that jointly promote compressibility, expressivity, and efficiency in the emergent language. First, interactive decoding incentivizes intermediate reasoning by requiring the receiver to produce partial reconstructions after each symbol. Second, a reconstruction-based imitation phase, inspired by iterated learning, trains successive generations of agents to imitate reconstructions rather than messages, enforcing a tighter communication bottleneck. Third, pairwise distance maximization regularizes message diversity by encouraging high distances between messages, with formal links to entropy maximization. Our method significantly improves both the efficiency and compositionality of the learned messages on the Shapes3D and MPI3D datasets, surpassing prior discrete communication frameworks in both reconstruction accuracy and topographic similarity. This work provides new theoretical and empirical evidence for the emergence of structured, generalizable communication protocols from simplicity-based inductive biases.

Vladimir Araujo, Andrés Carvallo, Carlos Aspillaga et al.

The success of pretrained word embeddings has motivated their use in the biomedical domain, with contextualized embeddings yielding remarkable results in several biomedical NLP tasks. However, there is a lack of research on quantifying their behavior under severe "stress" scenarios. In this work, we systematically evaluate three language models with adversarial examples -- automatically constructed tests that allow us to examine how robust the models are. We propose two types of stress scenarios focused on the biomedical named entity recognition (NER) task, one inspired by spelling errors and another based on the use of synonyms for medical terms. Our experiments with three benchmarks show that the performance of the original models decreases considerably, in addition to revealing their weaknesses and strengths. Finally, we show that adversarial training causes the models to improve their robustness and even to exceed the original performance in some cases.

Iván Cantador, Andrés Carvallo, Fernando Diez et al.

The success of neural network embeddings has entailed a renewed interest in using knowledge graphs for a wide variety of machine learning and information retrieval tasks. In particular, current recommendation methods based on graph embeddings have shown state-of-the-art performance. These methods commonly encode latent rating patterns and content features. Different from previous work, in this paper, we propose to exploit embeddings extracted from graphs that combine information from ratings and aspect-based opinions expressed in textual reviews. We then adapt and evaluate state-of-the-art graph embedding techniques over graphs generated from Amazon and Yelp reviews on six domains, outperforming baseline recommenders. Our approach has the advantage of providing explanations which leverage aspect-based opinions given by users about recommended items. Furthermore, we also provide examples of the applicability of recommendations utilizing aspect opinions as explanations in a visualization dashboard, which allows obtaining information about the most and least liked aspects of similar users obtained from the embeddings of an input graph.

Diego Rojo, Nyi Nyi Htun, Denis Parra et al.

Decision support systems have become increasingly popular in the domain of agriculture. With the development of automated machine learning, agricultural experts are now able to train, evaluate and make predictions using cutting edge machine learning (ML) models without the need for much ML knowledge. Although this automated approach has led to successful results in many scenarios, in certain cases (e.g., when few labeled datasets are available) choosing among different models with similar performance metrics is a difficult task. Furthermore, these systems do not commonly allow users to incorporate their domain knowledge that could facilitate the task of model selection, and to gain insight into the prediction system for eventual decision making. To address these issues, in this paper we present AHMoSe, a visual support system that allows domain experts to better understand, diagnose and compare different regression models, primarily by enriching model-agnostic explanations with domain knowledge. To validate AHMoSe, we describe a use case scenario in the viticulture domain, grape quality prediction, where the system enables users to diagnose and select prediction models that perform better. We also discuss feedback concerning the design of the tool from both ML and viticulture experts.

Andres Carvallo, Denis Parra, Gabriel Rada et al.

The COVID-19 has brought about a significant challenge to the whole of humanity, but with a special burden upon the medical community. Clinicians must keep updated continuously about symptoms, diagnoses, and effectiveness of emergent treatments under a never-ending flood of scientific literature. In this context, the role of evidence-based medicine (EBM) for curating the most substantial evidence to support public health and clinical practice turns essential but is being challenged as never before due to the high volume of research articles published and pre-prints posted daily. Artificial Intelligence can have a crucial role in this situation. In this article, we report the results of an applied research project to classify scientific articles to support Epistemonikos, one of the most active foundations worldwide conducting EBM. We test several methods, and the best one, based on the XLNet neural language model, improves the current approach by 93\% on average F1-score, saving valuable time from physicians who volunteer to curate COVID-19 research articles manually.

Pablo Pino, Denis Parra, Pablo Messina et al.

Several deep learning architectures have been proposed over the last years to deal with the problem of generating a written report given an imaging exam as input. Most works evaluate the generated reports using standard Natural Language Processing (NLP) metrics (e.g. BLEU, ROUGE), reporting significant progress. In this article, we contrast this progress by comparing state of the art (SOTA) models against weak baselines. We show that simple and even naive approaches yield near SOTA performance on most traditional NLP metrics. We conclude that evaluation methods in this task should be further studied towards correctly measuring clinical accuracy, ideally involving physicians to contribute to this end.

Pablo Messina, Pablo Pino, Denis Parra et al.

Every year physicians face an increasing demand of image-based diagnosis from patients, a problem that can be addressed with recent artificial intelligence methods. In this context, we survey works in the area of automatic report generation from medical images, with emphasis on methods using deep neural networks, with respect to: (1) Datasets, (2) Architecture Design, (3) Explainability and (4) Evaluation Metrics. Our survey identifies interesting developments, but also remaining challenges. Among them, the current evaluation of generated reports is especially weak, since it mostly relies on traditional Natural Language Processing (NLP) metrics, which do not accurately capture medical correctness.

Oleksii Moskalenko, Denis Parra, Diego Saez-Trumper

Wikipedia is edited by volunteer editors around the world. Considering the large amount of existing content (e.g. over 5M articles in English Wikipedia), deciding what to edit next can be difficult, both for experienced users that usually have a huge backlog of articles to prioritize, as well as for newcomers who that might need guidance in selecting the next article to contribute. Therefore, helping editors to find relevant articles should improve their performance and help in the retention of new editors. In this paper, we address the problem of recommending relevant articles to editors. To do this, we develop a scalable system on top of Graph Convolutional Networks and Doc2Vec, learning how to represent Wikipedia articles and deliver personalized recommendations for editors. We test our model on editors' histories, predicting their most recent edits based on their prior edits. We outperform competitive implicit-feedback collaborative-filtering methods such as WMRF based on ALS, as well as a traditional IR-method such as content-based filtering based on BM25. All of the data used on this paper is publicly available, including graph embeddings for Wikipedia articles, and we release our code to support replication of our experiments. Moreover, we contribute with a scalable implementation of a state-of-art graph embedding algorithm as current ones cannot efficiently handle the sheer size of the Wikipedia graph.

Pablo Messina, Manuel Cartagena, Patricio Cerda-Mardini et al.

Although there are several visually-aware recommendation models in domains like fashion or even movies, the art domain lacks thesame level of research attention, despite the recent growth of the online artwork market. To reduce this gap, in this article we introduceCuratorNet, a neural network architecture for visually-aware recommendation of art images. CuratorNet is designed at the core withthe goal of maximizing generalization: the network has a fixed set of parameters that only need to be trained once, and thereafter themodel is able to generalize to new users or items never seen before, without further training. This is achieved by leveraging visualcontent: items are mapped to item vectors through visual embeddings, and users are mapped to user vectors by aggregating the visualcontent of items they have consumed. Besides the model architecture, we also introduce novel triplet sampling strategies to build atraining set for rank learning in the art domain, resulting in more effective learning than naive random sampling. With an evaluationover a real-world dataset of physical paintings, we show that CuratorNet achieves the best performance among several baselines,including the state-of-the-art model VBPR. CuratorNet is motivated and evaluated in the art domain, but its architecture and trainingscheme could be adapted to recommend images in other areas

Vladimir Araujo, Andres Carvallo, Carlos Aspillaga et al.

The success of pre-trained word embeddings has motivated its use in tasks in the biomedical domain. The BERT language model has shown remarkable results on standard performance metrics in tasks such as Named Entity Recognition (NER) and Semantic Textual Similarity (STS), which has brought significant progress in the field of NLP. However, it is unclear whether these systems work seemingly well in critical domains, such as legal or medical. For that reason, in this work, we propose an adversarial evaluation scheme on two well-known datasets for medical NER and STS. We propose two types of attacks inspired by natural spelling errors and typos made by humans. We also propose another type of attack that uses synonyms of medical terms. Under these adversarial settings, the accuracy of the models drops significantly, and we quantify the extent of this performance loss. We also show that we can significantly improve the robustness of the models by training them with adversarial examples. We hope our work will motivate the use of adversarial examples to evaluate and develop models with increased robustness for medical tasks.

Felipe del Rio, Pablo Messina, Vicente Dominguez et al.

Visual embeddings from Convolutional Neural Networks (CNN) trained on the ImageNet dataset for the ILSVRC challenge have shown consistently good performance for transfer learning and are widely used in several tasks, including image recommendation. However, some important questions have not yet been answered in order to use these embeddings for a larger scope of recommendation domains: a) Do CNNs that perform better in ImageNet are also better for transfer learning in content-based image recommendation?, b) Does fine-tuning help to improve performance? and c) Which is the best way to perform the fine-tuning? In this paper we compare several CNN models pre-trained with ImageNet to evaluate their transfer learning performance to an artwork image recommendation task. Our results indicate that models with better performance in the ImageNet challenge do not always imply better transfer learning for recommendation tasks (e.g. NASNet vs. ResNet). Our results also show that fine-tuning can be helpful even with a small dataset, but not every fine-tuning works. Our results can inform other researchers and practitioners on how to train their CNNs for better transfer learning towards image recommendation systems.

Vicente Dominguez, Pablo Messina, Denis Parra et al.

Advances in image processing and computer vision in the latest years have brought about the use of visual features in artwork recommendation. Recent works have shown that visual features obtained from pre-trained deep neural networks (DNNs) perform very well for recommending digital art. Other recent works have shown that explicit visual features (EVF) based on attractiveness can perform well in preference prediction tasks, but no previous work has compared DNN features versus specific attractiveness-based visual features (e.g. brightness, texture) in terms of recommendation performance. In this work, we study and compare the performance of DNN and EVF features for the purpose of physical artwork recommendation using transactional data from UGallery, an online store of physical paintings. In addition, we perform an exploratory analysis to understand if DNN embedded features have some relation with certain EVF. Our results show that DNN features outperform EVF, that certain EVF features are more suited for physical artwork recommendation and, finally, we show evidence that certain neurons in the DNN might be partially encoding visual features such as brightness, providing an opportunity for explaining recommendations based on visual neural models.

Felipe del-Rio, Denis Parra, Jovan Kuzmicic et al.

Several studies indicate that attracting students to research careers requires to engage them from early undergraduate years. Following this paradigm, our Engineering School has developed an undergraduate research program that allows students to enroll in research in exchange for course credits. Moreover, we developed a web portal to inform students about the program and the opportunities, but participation remains lower than expected. In order to promote student engagement, we attempt to build a personalized recommender system of research opportunities to undergraduates. With this goal in mind we investigate two tasks. First, one that identifies students that are more willing to participate on this kind of program. A second task is generating a personalized list of recommendations of research opportunities for each student. To evaluate our approach, we perform a simulated prediction experiment with data from our School, which has more than 4,000 active undergraduate students nowadays. Our results indicate that there is a big potential to create a personalized recommender system for this purpose. Our results can be used as a baseline for colleges seeking strategies to encourage research activities within undergraduate students.

Gabriel Sepulveda, Vicente Dominguez, Denis Parra

This paper introduces pyRecLab, a software library written in C++ with Python bindings which allows to quickly train, test and develop recommender systems. Although there are several software libraries for this purpose, only a few let developers to get quickly started with the most traditional methods, permitting them to try different parameters and approach several tasks without a significant loss of performance. Among the few libraries that have all these features, they are available in languages such as Java, Scala or C#, what is a disadvantage for less experienced programmers more used to the popular Python programming language. In this article we introduce details of pyRecLab, showing as well performance analysis in terms of error metrics (MAE and RMSE) and train/test time. We benchmark it against the popular Java-based library LibRec, showing similar results. We expect programmers with little experience and people interested in quickly prototyping recommender systems to be benefited from pyRecLab.

Pablo Messina, Vicente Dominguez, Denis Parra et al.

Compared to other areas, artwork recommendation has received little attention, despite the continuous growth of the artwork market. Previous research has relied on ratings and metadata to make artwork recommendations, as well as visual features extracted with deep neural networks (DNN). However, these features have no direct interpretation to explicit visual features (e.g. brightness, texture) which might hinder explainability and user-acceptance. In this work, we study the impact of artwork metadata as well as visual features (DNN-based and attractiveness-based) for physical artwork recommendation, using images and transaction data from the UGallery online artwork store. Our results indicate that: (i) visual features perform better than manually curated data, (ii) DNN-based visual features perform better than attractiveness-based ones, and (iii) a hybrid approach improves the performance further. Our research can inform the development of new artwork recommenders relying on diverse content data.

Ivania Donoso, Denis Parra

Evidence-based health care (EBHC) is an important practice of medicine which attempts to provide systematic scientific evidence to answer clinical questions. In this context, Epistemonikos (www.epistemonikos.org) is one of the first and most important online systems in the field, providing an interface that supports users on searching and filtering scientific articles for practicing EBHC. The system nowadays requires a large amount of expert human effort, where close to 500 physicians manually curate articles to be utilized in the platform. In order to scale up the large and continuous amount of data to keep the system updated, we introduce EpistAid, an interactive intelligent interface which supports clinicians in the process of curating documents for Epistemonikos within lists of papers called evidence matrices. We introduce the characteristics, design and algorithms of our solution, as well as a prototype implementation and a case study to show how our solution addresses the information overload problem in this area.

Emanuel Lacic, Dominik Kowald, Paul Seitlinger et al.

In this work we present a novel item recommendation approach that aims at improving Collaborative Filtering (CF) in social tagging systems using the information about tags and time. Our algorithm follows a two-step approach, where in the first step a potentially interesting candidate item-set is found using user-based CF and in the second step this candidate item-set is ranked using item-based CF. Within this ranking step we integrate the information of tag usage and time using the Base-Level Learning (BLL) equation coming from human memory theory that is used to determine the reuse-probability of words and tags using a power-law forgetting function. As the results of our extensive evaluation conducted on data-sets gathered from three social tagging systems (BibSonomy, CiteULike and MovieLens) show, the usage of tag-based and time information via the BLL equation also helps to improve the ranking and recommendation process of items and thus, can be used to realize an effective item recommender that outperforms two alternative algorithms which also exploit time and tag-based information.

Emanuel Lacic, Dominik Kowald, Lukas Eberhard et al.

Recent research has unveiled the importance of online social networks for improving the quality of recommender systems and encouraged the research community to investigate better ways of exploiting the social information for recommendations. To contribute to this sparse field of research, in this paper we exploit users' interactions along three data sources (marketplace, social network and location-based) to assess their performance in a barely studied domain: recommending products and domains of interests (i.e., product categories) to people in an online marketplace environment. To that end we defined sets of content- and network-based user similarity features for each data source and studied them isolated using an user-based Collaborative Filtering (CF) approach and in combination via a hybrid recommender algorithm, to assess which one provides the best recommendation performance. Interestingly, in our experiments conducted on a rich dataset collected from SecondLife, a popular online virtual world, we found that recommenders relying on user similarity features obtained from the social network data clearly yielded the best results in terms of accuracy in case of predicting products, whereas the features obtained from the marketplace and location-based data sources also obtained very good results in case of predicting categories. This finding indicates that all three types of data sources are important and should be taken into account depending on the level of specialization of the recommendation task.