Elena Baralis

Giuseppe Attanasio, Debora Nozza, Dirk Hovy et al.

Natural Language Processing (NLP) models risk overfitting to specific terms in the training data, thereby reducing their performance, fairness, and generalizability. E.g., neural hate speech detection models are strongly influenced by identity terms like gay, or women, resulting in false positives, severe unintended bias, and lower performance. Most mitigation techniques use lists of identity terms or samples from the target domain during training. However, this approach requires a-priori knowledge and introduces further bias if important terms are neglected. Instead, we propose a knowledge-free Entropy-based Attention Regularization (EAR) to discourage overfitting to training-specific terms. An additional objective function penalizes tokens with low self-attention entropy. We fine-tune BERT via EAR: the resulting model matches or exceeds state-of-the-art performance for hate speech classification and bias metrics on three benchmark corpora in English and Italian. EAR also reveals overfitting terms, i.e., terms most likely to induce bias, to help identify their effect on the model, task, and predictions.

Alkis Koudounas, Moreno La Quatra, Lorenzo Vaiani et al.

Recent large-scale Spoken Language Understanding datasets focus predominantly on English and do not account for language-specific phenomena such as particular phonemes or words in different lects. We introduce ITALIC, the first large-scale speech dataset designed for intent classification in Italian. The dataset comprises 16,521 crowdsourced audio samples recorded by 70 speakers from various Italian regions and annotated with intent labels and additional metadata. We explore the versatility of ITALIC by evaluating current state-of-the-art speech and text models. Results on intent classification suggest that increasing scale and running language adaptation yield better speech models, monolingual text models outscore multilingual ones, and that speech recognition on ITALIC is more challenging than on existing Italian benchmarks. We release both the dataset and the annotation scheme to streamline the development of new Italian SLU models and language-specific datasets.

Eliana Pastor, Alkis Koudounas, Giuseppe Attanasio et al.

Recent advances in eXplainable AI (XAI) have provided new insights into how models for vision, language, and tabular data operate. However, few approaches exist for understanding speech models. Existing work focuses on a few spoken language understanding (SLU) tasks, and explanations are difficult to interpret for most users. We introduce a new approach to explain speech classification models. We generate easy-to-interpret explanations via input perturbation on two information levels. 1) Word-level explanations reveal how each word-related audio segment impacts the outcome. 2) Paralinguistic features (e.g., prosody and background noise) answer the counterfactual: ``What would the model prediction be if we edited the audio signal in this way?'' We validate our approach by explaining two state-of-the-art SLU models on two speech classification tasks in English and Italian. Our findings demonstrate that the explanations are faithful to the model's inner workings and plausible to humans. Our method and findings pave the way for future research on interpreting speech models.

Flavio Giobergia, Eliana Pastor, Luca de Alfaro et al.

Concept drift is a common phenomenon in data streams where the statistical properties of the target variable change over time. Traditionally, drift is assumed to occur globally, affecting the entire dataset uniformly. However, this assumption does not always hold true in real-world scenarios where only specific subpopulations within the data may experience drift. This paper explores the concept of localized drift and evaluates the performance of several drift detection techniques in identifying such localized changes. We introduce a synthetic dataset based on the Agrawal generator, where drift is induced in a randomly chosen subgroup. Our experiments demonstrate that commonly adopted drift detection methods may fail to detect drift when it is confined to a small subpopulation. We propose and test various drift detection approaches to quantify their effectiveness in this localized drift scenario. We make the source code for the generation of the synthetic benchmark available at https://github.com/fgiobergia/subgroup-agrawal-drift.

Daniele Rege Cambrin, Eleonora Poeta, Eliana Pastor et al.

Segmentation of crop fields is essential for enhancing agricultural productivity, monitoring crop health, and promoting sustainable practices. Deep learning models adopted for this task must ensure accurate and reliable predictions to avoid economic losses and environmental impact. The newly proposed Kolmogorov-Arnold networks (KANs) offer promising advancements in the performance of neural networks. This paper analyzes the integration of KAN layers into the U-Net architecture (U-KAN) to segment crop fields using Sentinel-2 and Sentinel-1 satellite images and provides an analysis of the performance and explainability of these networks. Our findings indicate a 2\% improvement in IoU compared to the traditional full-convolutional U-Net model in fewer GFLOPs. Furthermore, gradient-based explanation techniques show that U-KAN predictions are highly plausible and that the network has a very high ability to focus on the boundaries of cultivated areas rather than on the areas themselves. The per-channel relevance analysis also reveals that some channels are irrelevant to this task.

Moreno La Quatra, Alkis Koudounas, Lorenzo Vaiani et al. · gatech

Limited diversity in standardized benchmarks for evaluating audio representation learning (ARL) methods may hinder systematic comparison of current methods' capabilities. We present ARCH, a comprehensive benchmark for evaluating ARL methods on diverse audio classification domains, covering acoustic events, music, and speech. ARCH comprises 12 datasets, that allow us to thoroughly assess pre-trained SSL models of different sizes. ARCH streamlines benchmarking of ARL techniques through its unified access to a wide range of domains and its ability to readily incorporate new datasets and models. To address the current lack of open-source, pre-trained models for non-speech audio, we also release new pre-trained models that demonstrate strong performance on non-speech datasets. We argue that the presented wide-ranging evaluation provides valuable insights into state-of-the-art ARL methods, and is useful to pinpoint promising research directions.

Flavio Giobergia, Alkis Koudounas, Elena Baralis

Exploring exoplanets has transformed our understanding of the universe by revealing many planetary systems that defy our current understanding. To study their atmospheres, spectroscopic observations are used to infer essential atmospheric properties that are not directly measurable. Estimating atmospheric parameters that best fit the observed spectrum within a specified atmospheric model is a complex problem that is difficult to model. In this paper, we present a multi-target probabilistic regression approach that combines deep learning and inverse modeling techniques within a multimodal architecture to extract atmospheric parameters from exoplanets. Our methodology overcomes computational limitations and outperforms previous approaches, enabling efficient analysis of exoplanetary atmospheres. This research contributes to advancements in the field of exoplanet research and offers valuable insights for future studies.

Flavio Giobergia, Eliana Pastor, Luca de Alfaro et al.

The ability to detect and adapt to changes in data distributions is crucial to maintain the accuracy and reliability of machine learning models. Detection is generally approached by observing the drift of model performance from a global point of view. However, drifts occurring in (fine-grained) data subgroups may go unnoticed when monitoring global drift. We take a different perspective, and introduce methods for observing drift at the finer granularity of subgroups. Relevant data subgroups are identified during training and monitored efficiently throughout the model's life. Performance drifts in any subgroup are detected, quantified and characterized so as to provide an interpretable summary of the model behavior over time. Experimental results confirm that our subgroup-level drift analysis identifies drifts that do not show at the (coarser) global dataset level. The proposed approach provides a valuable tool for monitoring model performance in dynamic real-world applications, offering insights into the evolving nature of data and ultimately contributing to more robust and adaptive models.

Eleonora Poeta, Gabriele Ciravegna, Eliana Pastor et al.

The field of explainable artificial intelligence emerged in response to the growing need for more transparent and reliable models. However, using raw features to provide explanations has been disputed in several works lately, advocating for more user-understandable explanations. To address this issue, a wide range of papers proposing Concept-based eXplainable Artificial Intelligence (C-XAI) methods have arisen in recent years. Nevertheless, a unified categorization and precise field definition are still missing. This paper fills the gap by offering a thorough review of C-XAI approaches. We define and identify different concepts and explanation types. We provide a taxonomy identifying nine categories and propose guidelines for selecting a suitable category based on the development context. Additionally, we report common evaluation strategies including metrics, human evaluations and dataset employed, aiming to assist the development of future methods. We believe this survey will serve researchers, practitioners, and domain experts in comprehending and advancing this innovative field.

Alkis Koudounas, Claudio Savelli, Flavio Giobergia et al.

Machine unlearning, the process of efficiently removing specific information from machine learning models, is a growing area of interest for responsible AI. However, few studies have explored the effectiveness of unlearning methods on complex tasks, particularly speech-related ones. This paper introduces UnSLU-BENCH, the first benchmark for machine unlearning in spoken language understanding (SLU), focusing on four datasets spanning four languages. We address the unlearning of data from specific speakers as a way to evaluate the quality of potential "right to be forgotten" requests. We assess eight unlearning techniques and propose a novel metric to simultaneously better capture their efficacy, utility, and efficiency. UnSLU-BENCH sets a foundation for unlearning in SLU and reveals significant differences in the effectiveness and computational feasibility of various techniques.

Alkis Koudounas, Moreno La Quatra, Manuel Giollo et al. · gatech

Hallucinations in automatic speech recognition (ASR) systems refer to fluent and coherent transcriptions produced by neural ASR models that are completely unrelated to the underlying acoustic input (i.e., the speech signal). While similar to conventional decoding errors in potentially compromising the usability of transcriptions for downstream applications, hallucinations can be more detrimental due to their preservation of syntactically and semantically plausible structure. This apparent coherence can mislead subsequent processing stages and introduce serious risks, particularly in critical domains such as healthcare and law. Conventional evaluation metrics are primarily centered on error-based metrics and fail to distinguish between phonetic inaccuracies and hallucinations. Consequently, there is a critical need for new evaluation frameworks that can effectively identify and assess models with a heightened propensity for generating hallucinated content. To this end, we introduce SHALLOW, the first benchmark framework that systematically categorizes and quantifies hallucination phenomena in ASR along four complementary axes: lexical, phonetic, morphological, and semantic. We define targeted metrics within each category to produce interpretable profiles of model behavior. Through evaluation across various architectures and speech domains, we have found that SHALLOW metrics correlate strongly with word error rate (WER) when recognition quality is high (i.e., low WER). Still, this correlation weakens substantially as WER increases. SHALLOW, therefore, captures fine-grained error patterns that WER fails to distinguish under degraded and challenging conditions. Our framework supports specific diagnosis of model weaknesses and provides feedback for model improvement beyond what aggregate error rates can offer.

Daniele Rege Cambrin, Eleonora Poeta, Eliana Pastor et al.

Forecasting surface water dynamics is crucial for water resource management and climate change adaptation. However, the field lacks comprehensive datasets and standardized benchmarks. In this paper, we introduce HydroChronos, a large-scale, multi-modal spatiotemporal dataset for surface water dynamics forecasting designed to address this gap. We couple the dataset with three forecasting tasks. The dataset includes over three decades of aligned Landsat 5 and Sentinel-2 imagery, climate data, and Digital Elevation Models for diverse lakes and rivers across Europe, North America, and South America. We also propose AquaClimaTempo UNet, a novel spatiotemporal architecture with a dedicated climate data branch, as a strong benchmark baseline. Our model significantly outperforms a Persistence baseline for forecasting future water dynamics by +14% and +11% F1 across change detection and direction of change classification tasks, and by +0.1 MAE on the magnitude of change regression. Finally, we conduct an Explainable AI analysis to identify the key climate variables and input channels that influence surface water change, providing insights to inform and guide future modeling efforts.

Moreno La Quatra, Alkis Koudounas, Elena Baralis et al.

Italy exhibits rich linguistic diversity across its territory due to the distinct regional languages spoken in different areas. Recent advances in self-supervised learning provide new opportunities to analyze Italy's linguistic varieties using speech data alone. This includes the potential to leverage representations learned from large amounts of data to better examine nuances between closely related linguistic varieties. In this study, we focus on automatically identifying the geographic region of origin of speech samples drawn from Italy's diverse language varieties. We leverage self-supervised learning models to tackle this task and analyze differences and similarities between Italy's regional languages. In doing so, we also seek to uncover new insights into the relationships among these diverse yet closely related varieties, which may help linguists understand their interconnected evolution and regional development over time and space. To improve the discriminative ability of learned representations, we evaluate several supervised contrastive learning objectives, both as pre-training steps and additional fine-tuning objectives. Experimental evidence shows that pre-trained self-supervised models can effectively identify regions from speech recording. Additionally, incorporating contrastive objectives during fine-tuning improves classification accuracy and yields embeddings that distinctly separate regional varieties, demonstrating the value of combining self-supervised pre-training and contrastive learning for this task.

Alkis Koudounas, Gabriele Ciravegna, Marco Fantini et al.

Voice disorders are pathologies significantly affecting patient quality of life. However, non-invasive automated diagnosis of these pathologies is still under-explored, due to both a shortage of pathological voice data, and diversity of the recording types used for the diagnosis. This paper proposes a novel solution that adopts transformers directly working on raw voice signals and addresses data shortage through synthetic data generation and data augmentation. Further, we consider many recording types at the same time, such as sentence reading and sustained vowel emission, by employing a Mixture of Expert ensemble to align the predictions on different data types. The experimental results, obtained on both public and private datasets, show the effectiveness of our solution in the disorder detection and classification tasks and largely improve over existing approaches.

Alkis Koudounas, Flavio Giobergia, Eliana Pastor et al.

Speech models may be affected by performance imbalance in different population subgroups, raising concerns about fair treatment across these groups. Prior attempts to mitigate unfairness either focus on user-defined subgroups, potentially overlooking other affected subgroups, or do not explicitly improve the internal representation at the subgroup level. This paper proposes the first adoption of contrastive learning to mitigate speech model bias in underperforming subgroups. We employ a three-level learning technique that guides the model in focusing on different scopes for the contrastive loss, i.e., task, subgroup, and the errors within subgroups. The experiments on two spoken language understanding datasets and two languages demonstrate that our approach improves internal subgroup representations, thus reducing model bias and enhancing performance.

Eleonora Poeta, Flavio Giobergia, Eliana Pastor et al.

Kolmogorov-Arnold Networks (KANs) have very recently been introduced into the world of machine learning, quickly capturing the attention of the entire community. However, KANs have mostly been tested for approximating complex functions or processing synthetic data, while a test on real-world tabular datasets is currently lacking. In this paper, we present a benchmarking study comparing KANs and Multi-Layer Perceptrons (MLPs) on tabular datasets. The study evaluates task performance and training times. From the results obtained on the various datasets, KANs demonstrate superior or comparable accuracy and F1 scores, excelling particularly in datasets with numerous instances, suggesting robust handling of complex data. We also highlight that this performance improvement of KANs comes with a higher computational cost when compared to MLPs of comparable sizes.

Eliana Pastor, Luca de Alfaro, Elena Baralis

When analyzing the behavior of machine learning algorithms, it is important to identify specific data subgroups for which the considered algorithm shows different performance with respect to the entire dataset. The intervention of domain experts is normally required to identify relevant attributes that define these subgroups. We introduce the notion of divergence to measure this performance difference and we exploit it in the context of (i) classification models and (ii) ranking applications to automatically detect data subgroups showing a significant deviation in their behavior. Furthermore, we quantify the contribution of all attributes in the data subgroup to the divergent behavior by means of Shapley values, thus allowing the identification of the most impacting attributes.

Tania Cerquitelli, Stefano Proto, Francesco Ventura et al.

A key aspect of automating predictive machine learning entails the capability of properly triggering the update of the trained model. To this aim, suitable automatic solutions to self-assess the prediction quality and the data distribution drift between the original training set and the new data have to be devised. In this paper, we propose a novel methodology to automatically detect prediction-quality degradation of machine learning models due to class-based concept drift, i.e., when new data contains samples that do not fit the set of class labels known by the currently-trained predictive model. Experiments on synthetic and real-world public datasets show the effectiveness of the proposed methodology in automatically detecting and describing concept drift caused by changes in the class-label data distributions.

Luca Venturini, Elena Baralis, Paolo Garza

Supervised learning algorithms are nowadays successfully scaling up to datasets that are very large in volume, leveraging the potential of in-memory cluster-computing Big Data frameworks. Still, massive datasets with a number of large-domain categorical features are a difficult challenge for any classifier. Most off-the-shelf solutions cannot cope with this problem. In this work we introduce DAC, a Distributed Associative Classifier. DAC exploits ensemble learning to distribute the training of an associative classifier among parallel workers and improve the final quality of the model. Furthermore, it adopts several novel techniques to reach high scalability without sacrificing quality, among which a preventive pruning of classification rules in the extraction phase based on Gini impurity. We ran experiments on Apache Spark, on a real large-scale dataset with more than 4 billion records and 800 million distinct categories. The results showed that DAC improves on a state-of-the-art solution in both prediction quality and execution time. Since the generated model is human-readable, it can not only classify new records, but also allow understanding both the logic behind the prediction and the properties of the model, becoming a useful aid for decision makers.