Daniele Tessera

Prof. Daniele Tessera

Dipartimento di Matematica e Fisica, Università Cattolica del Sacro Cuore – Brescia, Brescia, Italy

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 May 2022 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Enrico BarbieratoEnrico Barbierato
Enrico Barbierato
MVMarco L. Della Vedova
Marco L. Della Vedova
DTDaniele Tessera
Daniele Tessera
Daniele TotiDaniele Toti
Daniele Toti
NVNicola Vanoli
Nicola Vanoli
The development of algorithms, based on machine learning techniques, supporting (or even replacing) human judgment must take into account concepts such as data bias and fairness. Though scientific literature proposes numerous techniques to detect and evaluate these problems, less attention has been dedicated to methods generating intentionally biased datasets, which could be used by data scientists to develop and validate unbiased and fair decision-making algorithms. To this end, this paper presents a novel method to generate a synthetic dataset, where bias can be modeled by using a probabilistic network exploiting structural equation modeling. The proposed methodology has been validated on a simple dataset to highlight the impact of tuning parameters on bias and fairness, as well as on a more realistic example based on a loan approval status dataset. In particular, this methodology requires a limited number of parameters compared to other techniques for generating datasets with a controlled amount of bias and fairness.

24 March 2025 in Neural Computing and Applications
Verified
Journal Article
Open access
PDF file
Andrea PozziAndrea Pozzi
Andrea Pozzi
AIAlessandro Incremona
Alessandro Incremona
DTDaniele Tessera
Daniele Tessera
Daniele TotiDaniele Toti
Daniele Toti
Knowledge distillation is recognized as a valuable model compression strategy that alleviates the computational burden of large language models while preserving performance. This strategy involves training a smaller model utilizing both real data and predictions from a more cumbersome model. Traditional distillation methods, however, are often compromised by exposure bias, which results from reliance on next-step prediction training loss. This bias emerges when models are tested in free-running mode, differing from their training regime and leading to a progressive drift in input distributions between testing and training phases. An analogous issue, known as ‘distributional shift’, has been effectively addressed in imitation learning through various methodologies. Therefore, this paper specifically tailors an imitation learning-based solution to a traditional knowledge distillation framework which inherently considers both real data and the teacher’s predictions as dual sources of expert demonstrations. The effectiveness of this approach is demonstrated over five different test datasets, where it outperforms traditional benchmarks across all evaluation metrics. Specifically, it achieves superior results in perplexity, multi-token generation, and G-Eval score, indicating improvements in both predictive accuracy and alignment with human judgment in text quality. These results underscore the potential of this approach to effectively address exposure bias in large language model distillation.

21 June 2021 in IEEE Access
Verified
Journal Article
Open access
MCMaria Carla Calzarossa
Maria Carla Calzarossa
MVMarco L. Della Vedova
Marco L. Della Vedova
LMLuisa Massari
Luisa Massari
GNGiuseppe Nebbione
Giuseppe Nebbione
DTDaniele Tessera
Daniele Tessera
Cloud technologies are being used nowadays to cope with the increased computing and storage requirements of services and applications. Nevertheless, decisions about resources to be provisioned and the corresponding scheduling plans are far from being easily made especially because of the variability and uncertainty affecting workload demands as well as technological infrastructure performance. In this paper we address these issues by formulating a multi-objective constrained optimization problem aimed at identifying the optimal scheduling plans for scientific workflows to be deployed in uncertain cloud environments. In particular, we focus on minimizing the expected workflow execution time and monetary cost under probabilistic constraints on deadline and budget. According to the proposed approach, this problem is solved offline, that is, prior to workflow execution, with the intention of allowing cloud users to choose the plan of the Pareto optimal set satisfying their requirements and preferences. The analysis of the combined effects of cloud uncertainty and probabilistic constraints has shown that the solutions of the optimization problem are strongly affected by uncertainty. Hence, to properly provision cloud resources, it is compelling to precisely quantify uncertainty and take explicitly into account its effects in the decision process.

7 April 2023 in IEEE Access
Verified
Journal Article
Open access
Manuele BettelliManuele Bettelli
Manuele Bettelli
FVFilippo Vurro
Filippo Vurro
Riccardo PecoriRiccardo Pecori
Riccardo Pecori
MIchela JanniMIchela Janni
MIchela Janni
NCNicola Coppedè
Nicola Coppedè
DTDaniele Tessera
Daniele Tessera
Water stress and in particular drought are some of the most significant factors affecting plant growth, food production, and thus food security. Furthermore, the possibility to predict and shape irrigation on real plant demands is priceless. The objective of this study is to characterize, classify, and forecast water stress in tomato plants by means of in vivo real time data obtained through a novel sensor, named bioristor, and of different artificial intelligence models. First of all, we have applied classification models, namely Decision Trees and Random Forest, to try to distinguish four different stress statuses of tomato plants. Then we have predicted, through the help of recurrent neural networks, the future status of a plant when considering both a binary (water stressed and not water stressed) and a four-status scenario. The obtained results are very good in terms of accuracy, precision, recall, f-measure, and of the resulting confusion matrices, and they suggest that the considered novel data and features coming from the bioristor, together with the used AI models, can be successfully applied, in the future, to a real-world on-the-field smart irrigation scenario.

22 December 2025 in Neurocomputing
Verified
Journal Article
Open access
AIAlessandro Incremona
Alessandro Incremona
Andrea PozziAndrea Pozzi
Andrea Pozzi
AGAlessandro Guiscardi
Alessandro Guiscardi
DTDaniele Tessera
Daniele Tessera

26 April 2024 in IEEE Access
Verified
Journal Article
Open access
FCFederico Castelletti
Federico Castelletti
FNFabrizio Niro
Fabrizio Niro
MDMarco Denti
Marco Denti
DTDaniele Tessera
Daniele Tessera
Andrea PozziAndrea Pozzi
Andrea Pozzi
Distributed energy generation systems, key for producing electricity near usage points, are essential to meet theglobal electricity demand, leveraging diverse sources likerenewables, traditional fuels, and industrial waste heat. Despite theirhigh reliability, these systemsare not immune to faults and failures. Such incidents can result in considerable downtime and reduced efficiency, underlining the need for effectivefault detection and diagnosis techniques. Implementing these strategies is crucial not just for mitigating damage and preventing potential disasters, but also to maintain optimal performance levels. This paper introduces a novel methodology based on Bayesian graphical modeling for unsupervised fault diagnosis, focusing on organic Rankine cycle case study. It employsstructural learning to discern unknown intervention points within a directed acyclic graph that models the power plant’s operations. By analyzing real-world data, the study demonstrates the effectiveness of this approach, pinpointing a subset of variables that could be implicated in specific faults.