arXiv subjects

Marco Brambilla

Publications and source records attributed to Marco Brambilla.

At least 19 recordsLinked to original sources

arXiv2026

A Framework for Evaluating Zero-Shot Image Generation in Concept-based Explainability

Concept-based Explainable Artificial Intelligence (XAI) interprets deep learning models using human-understandable visual features (e.g., textures or object parts) by linking internal representations to class predictions, thereby bridging the gap between low-level image data and high-level semantics. A major challenge, however, is the reliance on large sets of labeled images to represent each concept, which limits scalability. In this work, we investigate the use of zero-shot Text-to-Image (T2I) generative models as a source of synthetic concept datasets for concept-based XAI methods. Specifically, we generate concepts using predefined prompts and evaluate their faithfulness to real ones through four complementary analyses: (1) comparing synthetic vs. real concept images via concept representation similarity; (2) evaluating their intra-similarity by comparing pairs of subsets of the same concept with progressively increasing size; (3) evaluating their performance for downstream explanation tasks using relevant class images; (4) evaluating how removing a concept from tested class images affects explanations of generated concepts. While current T2I generative models promise a shortcut to concept-based XAI, our study highlights challenges and raises open questions about the use of synthetic data generated by zero-shot pipelines in model analyses. The resulting dataset is available at https://github.com/DataSciencePolimi/ZeroShot-T2I-Concepts.

cs.CV↗

arXiv2026

Comparing Human and Large Language Model Interpretation of Implicit Information

The interpretation of implicit meanings is an integral aspect of human communication. However, this framework may not transfer to interactions with Large Language Models (LLMs). To investigate this, we introduce the task of Implicit Information Extraction (IIE) and propose an LLM-based IIE pipeline that builds a structured knowledge graph from a context sentence by extracting relational triplets, validating implicit inferences, and analyzing temporal relations. We evaluate two LLMs against crowdsourced human judgments on two datasets. We find that humans agree with most model triplets yet consistently propose many additions, indicating limited coverage in current LLM-based IIE. Moreover, in our experiments, models appear to be more conservative about implicit inferences than humans in socially rich contexts, whereas humans become more conservative in shorter, fact-oriented contexts. Our code is available at https://github.com/Antonio-Dee/IIE_from_LLM.

cs.CL↗

arXiv2026

LLM-Enhanced Semantic Data Integration of Electronic Component Qualifications in the Aerospace Domain

Large manufacturing companies face challenges in information retrieval due to data silos maintained by different departments, leading to inconsistencies and misalignment across databases. This paper presents an experience in integrating and retrieving qualification data for electronic components used in satellite board design. Due to data silos, designers cannot immediately determine the qualification status of individual components. However, this process is critical during the planning phase, when assembly drawings are issued before production, to optimize new qualifications and avoid redundant efforts. To address this, we propose a pipeline that uses Virtual Knowledge Graphs for a unified view over heterogeneous data sources and LLMs to enhance retrieval and reduce manual effort in data cleansing. The retrieval of qualifications is then performed through an Ontology-based Data Access approach for structured queries and a vector search mechanism for retrieving qualifications based on similar textual properties. We perform a comparative cost-benefit analysis, demonstrating that the proposed pipeline also outperforms approaches relying solely on LLMs, such as Retrieval-Augmented Generation (RAG), in terms of long-term efficiency.

cs.IR↗

arXiv2026

MDER-DR: Multi-Hop Question Answering with Entity-Centric Summaries

Retrieval-Augmented Generation (RAG) over Knowledge Graphs (KGs) suffers from the fact that indexing approaches may lose important contextual nuance when text is reduced to triples, thereby degrading performance in downstream Question-Answering (QA) tasks, particularly for multi-hop QA, which requires composing answers from multiple entities, facts, or relations. We propose a domain-agnostic, KG-based QA framework that covers both the indexing and retrieval/inference phases. A new indexing approach called Map-Disambiguate-Enrich-Reduce (MDER) generates context-derived triple descriptions and subsequently integrates them with entity-level summaries, thus avoiding the need for explicit traversal of edges in the graph during the QA retrieval phase. Complementing this, we introduce Decompose-Resolve (DR), a retrieval mechanism that decomposes user queries into resolvable triples and grounds them in the KG via iterative reasoning. Together, MDER and DR form an LLM-driven QA pipeline that is robust to sparse, incomplete, and complex relational data. Experiments show that on standard and domain specific benchmarks, MDER-DR achieves substantial improvements over standard RAG baselines (up to 66%), while maintaining cross-lingual robustness. Our code is available at https://github.com/DataSciencePolimi/MDER-DR_RAG.

cs.CL↗

arXiv2026

Learning Concept Bottleneck Models from Mechanistic Explanations

Concept Bottleneck Models (CBMs) aim for ante-hoc interpretability by learning a bottleneck layer that predicts interpretable concepts before the decision. State-of-the-art approaches typically select which concepts to learn via human specification, open knowledge graphs, prompting an LLM, or using general CLIP concepts. However, concepts defined a-priori may not have sufficient predictive power for the task or even be learnable from the available data. As a result, these CBMs often significantly trail their black-box counterpart when controlling for information leakage. To address this, we introduce a novel CBM pipeline named Mechanistic CBM (M-CBM), which builds the bottleneck directly from a black-box model's own learned concepts. These concepts are extracted via Sparse Autoencoders (SAEs) and subsequently named and annotated on a selected subset of images using a Multimodal LLM. For fair comparison and leakage control, we also introduce the Number of Contributing Concepts (NCC), a decision-level sparsity metric that extends the recently proposed NEC metric. Across diverse datasets, we show that M-CBMs consistently surpass prior CBMs at matched sparsity, while improving concept predictions and providing concise explanations. Our code is available at https://github.com/Antonio-Dee/M-CBM.

cs.LG↗

arXiv2025

A Graph-based RAG for Energy Efficiency Question Answering

In this work, we investigate the use of Large Language Models (LLMs) within a graph-based Retrieval Augmented Generation (RAG) architecture for Energy Efficiency (EE) Question Answering. First, the system automatically extracts a Knowledge Graph (KG) from guidance and regulatory documents in the energy field. Then, the generated graph is navigated and reasoned upon to provide users with accurate answers in multiple languages. We implement a human-based validation using the RAGAs framework properties, a validation dataset comprising 101 question-answer pairs, and domain experts. Results confirm the potential of this architecture and identify its strengths and weaknesses. Validation results show how the system correctly answers in about three out of four of the cases (75.2 +- 2.7%), with higher results on questions related to more general EE answers (up to 81.0 +- 4.1%), and featuring promising multilingual abilities (4.4% accuracy loss due to translation).

cs.CL↗

arXiv2025

A data-driven analysis of the impact of non-compliant individuals on epidemic diffusion in urban settings

Individuals who do not comply with public health safety measures pose a significant challenge to effective epidemic control, as their risky behaviours can undermine public health interventions. This is particularly relevant in urban environments because of their high population density and complex social interactions. In this study, we employ detailed contact networks, built using a data-driven approach, to examine the impact of non-compliant individuals on epidemic dynamics in three major Italian cities: Torino, Milano, and Palermo. We use a heterogeneous extension of the Susceptible-Infected-Recovered model that distinguishes between ordinary and non-compliant individuals, who are more infectious and/or more susceptible. By combining electoral data with recent findings on vaccine hesitancy, we obtain spatially heterogeneous distributions of non-compliance. Epidemic simulations demonstrate that even a small proportion of non-compliant individuals in the population can substantially increase the number of infections and accelerate the timing of their peak. Furthermore, the impact of non-compliance is greatest when disease transmission rates are moderate. Including the heterogeneous, data-driven distribution of non-compliance in the simulations results in infection hotspots forming with varying intensity according to the disease transmission rate. Overall, these findings emphasise the importance of monitoring behavioural compliance and tailoring public health interventions to address localised risks.

physics.soc-ph↗

arXiv2024

Visual-TCAV: Concept-based Attribution and Saliency Maps for Post-hoc Explainability in Image Classification

Convolutional Neural Networks (CNNs) have shown remarkable performance in image classification. However, interpreting their predictions is challenging due to the size and complexity of these models. State-of-the-art saliency methods generate local explanations highlighting the area in the input image where a class is identified but cannot explain how a concept of interest contributes to the prediction. On the other hand, concept-based methods, such as TCAV, provide insights into how sensitive the network is to a human-defined concept but cannot compute its attribution in a specific prediction nor show its location within the input image. We introduce Visual-TCAV, a novel explainability framework aiming to bridge the gap between these methods by providing both local and global explanations. Visual-TCAV uses Concept Activation Vectors (CAVs) to generate class-agnostic saliency maps that show where the network recognizes a certain concept. Moreover, it can estimate the attribution of these concepts to the output of any class using a generalization of Integrated Gradients. We evaluate the method's faithfulness via a controlled experiment where the ground truth for explanations is known, showing better ground truth alignment than TCAV. Our code is available at https://github.com/DataSciencePolimi/Visual-TCAV.

cs.CV↗

arXiv2024

Integrating Large Language Models and Knowledge Graphs for Extraction and Validation of Textual Test Data

Aerospace manufacturing companies, such as Thales Alenia Space, design, develop, integrate, verify, and validate products characterized by high complexity and low volume. They carefully document all phases for each product but analyses across products are challenging due to the heterogeneity and unstructured nature of the data in documents. In this paper, we propose a hybrid methodology that leverages Knowledge Graphs (KGs) in conjunction with Large Language Models (LLMs) to extract and validate data contained in these documents. We consider a case study focused on test data related to electronic boards for satellites. To do so, we extend the Semantic Sensor Network ontology. We store the metadata of the reports in a KG, while the actual test results are stored in parquet accessible via a Virtual Knowledge Graph. The validation process is managed using an LLM-based approach. We also conduct a benchmarking study to evaluate the performance of state-of-the-art LLMs in executing this task. Finally, we analyze the costs and benefits of automating preexisting processes of manual data extraction and validation for subsequent cross-report analyses.

cs.AI↗

arXiv2024

Interpretable Network Visualizations: A Human-in-the-Loop Approach for Post-hoc Explainability of CNN-based Image Classification

Transparency and explainability in image classification are essential for establishing trust in machine learning models and detecting biases and errors. State-of-the-art explainability methods generate saliency maps to show where a specific class is identified, without providing a detailed explanation of the model's decision process. Striving to address such a need, we introduce a post-hoc method that explains the entire feature extraction process of a Convolutional Neural Network. These explanations include a layer-wise representation of the features the model extracts from the input. Such features are represented as saliency maps generated by clustering and merging similar feature maps, to which we associate a weight derived by generalizing Grad-CAM for the proposed methodology. To further enhance these explanations, we include a set of textual labels collected through a gamified crowdsourcing activity and processed using NLP techniques and Sentence-BERT. Finally, we show an approach to generate global explanations by aggregating labels across multiple images.

cs.LG↗

arXiv2023

My Lockdown Escape: Sparking Self-Empathy in the Context of the Covid-19 Pandemic

During the Covid-19 pandemic, research communities focused on collecting and understanding people's behaviours and feelings to study and tackle the pandemic indirect effects. Despite its consequences are slowly starting to fade away, such an interest is still alive. In this article, we propose a hybrid, gamified, story-driven data collection approach to spark self-empathy, hence resurfacing people's past feelings. The game is designed to include a physical board, decks of cards, and a digital application. As the player plays through the game, they customize and escape from their lockdown room by completing statements and answering a series of questions that define their story. The decoration of the lockdown room and the storytelling-driven approach are targeted at sparking people's emotions and self-empathy towards their past selves. Ultimately, the proposed approach was proven effective in sparking and collecting feelings, while a few improvements are still necessary.

cs.HC↗

arXiv2022

A.I. Robustness: a Human-Centered Perspective on Technological Challenges and Opportunities

Despite the impressive performance of Artificial Intelligence (AI) systems, their robustness remains elusive and constitutes a key issue that impedes large-scale adoption. Robustness has been studied in many domains of AI, yet with different interpretations across domains and contexts. In this work, we systematically survey the recent progress to provide a reconciled terminology of concepts around AI robustness. We introduce three taxonomies to organize and describe the literature both from a fundamental and applied point of view: 1) robustness by methods and approaches in different phases of the machine learning pipeline; 2) robustness for specific model architectures, tasks, and systems; and in addition, 3) robustness assessment methodologies and insights, particularly the trade-offs with other trustworthiness properties. Finally, we identify and discuss research gaps and opportunities and give an outlook on the field. We highlight the central role of humans in evaluating and enhancing AI robustness, considering the necessary knowledge humans can provide, and discuss the need for better understanding practices and developing supportive tools in the future.

cs.AI↗

arXiv2022

COCTEAU: an Empathy-Based Tool for Decision-Making

Traditional approaches to data-informed policymaking are often tailored to specific contexts and lack strong citizen involvement and collaboration, which are required to design sustainable policies. We argue the importance of empathy-based methods in the policymaking domain given the successes in diverse settings, such as healthcare and education. In this paper, we introduce COCTEAU (Co-Creating The European Union), a novel framework built on the combination of empathy and gamification to create a tool aimed at strengthening interactions between citizens and policy-makers. We describe our design process and our concrete implementation, which has already undergone preliminary assessments with different stakeholders. Moreover, we briefly report pilot results from the assessment. Finally, we describe the structure and goals of our demonstration regarding the newfound formats and organizational aspects of academic conferences.

cs.HC↗

arXiv2022

Empathy-Centric Design At Scale

EmpathiCH aims at bringing together and blend different expertise to develop new research agenda in the context of "Empathy-Centric Design at Scale". The main research question is to investigate how new technologies can contribute to the elicitation of empathy across and within multiple stakeholders at scale; and how empathy can be used to design solutions to societal problems that are not only effective but also balanced, inclusive, and aware of their effect on society. Through presentations, participatory sessions, and a living experiment -- where data about the peoples' interactions is collected throughout the event -- we aim to make this workshop the ideal venue to foster collaboration, build networks, and shape the future direction of "Empathy-Centric Design at Scale".

cs.HC↗

arXiv2022

VaccinEU: COVID-19 vaccine conversations on Twitter in French, German and Italian

Despite the increasing limitations for unvaccinated people, in many European countries there is still a non-negligible fraction of individuals who refuse to get vaccinated against SARS-CoV-2, undermining governmental efforts to eradicate the virus. We study the role of online social media in influencing individuals' opinion towards getting vaccinated by designing a large-scale collection of Twitter messages in three different languages -- French, German and Italian -- and providing public access to the data collected. Focusing on the European context, our VaccinEU dataset aims to help researchers to better understand the impact of online (mis)information about vaccines and design more accurate communication strategies to maximize vaccination coverage.

cs.SI↗

arXiv2021

Exploiting Twitter as Source of Large Corpora of Weakly Similar Pairs for Semantic Sentence Embeddings

Semantic sentence embeddings are usually supervisedly built minimizing distances between pairs of embeddings of sentences labelled as semantically similar by annotators. Since big labelled datasets are rare, in particular for non-English languages, and expensive, recent studies focus on unsupervised approaches that require not-paired input sentences. We instead propose a language-independent approach to build large datasets of pairs of informal texts weakly similar, without manual human effort, exploiting Twitter's intrinsic powerful signals of relatedness: replies and quotes of tweets. We use the collected pairs to train a Transformer model with triplet-like structures, and we test the generated embeddings on Twitter NLP similarity tasks (PIT and TURL) and STSb. We also introduce four new sentence ranking evaluation benchmarks of informal texts, carefully extracted from the initial collections of tweets, proving not only that our best model learns classical Semantic Textual Similarity, but also excels on tasks where pairs of sentences are not exact paraphrases. Ablation studies reveal how increasing the corpus size influences positively the results, even at 2M samples, suggesting that bigger collections of Tweets still do not contain redundant information about semantic similarities.

cs.CL↗

arXiv2021

VaccinItaly: monitoring Italian conversations around vaccines on Twitter and Facebook

We present VaccinItaly, a project which monitors Italian online conversations around vaccines, on Twitter and Facebook. We describe the ongoing data collection, which follows the SARS-CoV-2 vaccination campaign roll-out in Italy and we provide public access to the data collected. We show results from a preliminary analysis of the spread of low- and high-credibility news shared alongside vaccine-related conversations on both social media platforms. We also investigate the content of most popular YouTube videos and encounter several cases of harmful and misleading content about vaccines. Finally, we geolocate Twitter users who discuss vaccines and correlate their activity with open data statistics on vaccine uptake. We make up-to-date results available to the public through an interactive online dashboard associated with the project. The goal of our project is to gain further understanding of the interplay between the public discourse on online social media and the dynamics of vaccine uptake in the real world.

cs.SI↗

arXiv2020

EFSG: Evolutionary Fooling Sentences Generator

Large pre-trained language representation models (LMs) have recently collected a huge number of successes in many NLP tasks. In 2018 BERT, and later its successors (e.g. RoBERTa), obtained state-of-the-art results in classical benchmark tasks, such as GLUE benchmark. After that, works about adversarial attacks have been published to test their generalization proprieties and robustness. In this work, we design Evolutionary Fooling Sentences Generator (EFSG), a model- and task-agnostic adversarial attack algorithm built using an evolutionary approach to generate false-positive sentences for binary classification tasks. We successfully apply EFSG to CoLA and MRPC tasks, on BERT and RoBERTa, comparing performances. Results prove the presence of weak spots in state-of-the-art LMs. We finally test adversarial training as a data augmentation defence approach against EFSG, obtaining stronger improved models with no loss of accuracy when tested on the original datasets.

cs.CL↗