When machines learn to remember.
In Proust’s novel, the taste of a madeleine unlocks an entire world buried in memory. MADELEinE (MAchine and DEep LEarning in Experiments) pursues an analogous ambition: teaching machines to discover hidden patterns in vast scientific datasets — the faint signals that would otherwise remain buried beneath noise.
MADELEinE is a research program (PI: Antonio Pagliaro) at INAF–IASF Palermo that applies Artificial Intelligence to astrophysics, astroparticle physics, quantitative finance, and computer vision. Our work spans the full AI pipeline: from classical ensemble methods (Random Forest, Extra Trees, XGBoost) to deep neural architectures for image classification, event reconstruction, and time series prediction.
Core research
Cherenkov astronomy. Our primary focus is the ASTRI Mini-Array — a next-generation array of Imaging Atmospheric Cherenkov Telescopes. We develop ML methods for gamma/hadron discrimination and energy reconstruction, achieving results that led to a cover story in Applied Sciences (Vol. 15, Issue 7). A stacking ensemble of Extra Trees, Random Forest and XGBoost has proven the most sensitive technique for signal segregation from the cosmic-ray background.
Data-driven finance. We apply the same rigorous ML methodology to financial markets — regime-aware forecasting with LightGBM conditioned on Hidden Markov Models, cognitive bias analysis in asset pricing, and critical reassessment of AI predictive power against the efficient market hypothesis.
Computer vision & atmospheric monitoring. Deep learning methods for cloud masking in multispectral satellite imagery, and infrared/UV data analysis from stratospheric balloons (EUSO-SPB1/SPB2) and space (Mini-EUSO) within the JEM-EUSO program.
Explainable AI. Across all domains, we pursue interpretability: not just accurate predictions, but understanding why models reach their conclusions — a requirement for trustworthy science.
Selected publications
High Energy Astrophysics
- Machine Learning-Enhanced Discrimination of Gamma-Ray and Hadron Events Using Temporal Features: An ASTRI Mini-Array Analysis — cover story, Appl. Sci. Vol. 15(7)
- Application of Machine Learning Ensemble Methods to ASTRI Mini-Array Cherenkov Event Reconstruction
- Application of Machine and Deep Learning Methods to the Analysis of IACTs Data
Data Driven Finance
- Regime-Aware LightGBM for Stock Market Forecasting: A Validated Walk-Forward Framework with Statistical Rigor and Explainable AI Analysis
- Cognitive Biases in Asset Pricing: An Empirical Analysis of the Alphabet Effect and Ticker Fluency in the US Market
- Artificial Intelligence vs. Efficient Markets: A Critical Reassessment of Predictive Models in the Big Data Era
- Forecasting Significant Stock Market Price Changes Using Machine Learning: Extra Trees Classifier Leads
- An Introduction to Machine Learning Methods for Fraud Detection
Computer Vision & More
- An Introduction to Machine and Deep Learning Methods for Cloud Masking Applications
- Advanced AI and Machine Learning Techniques for Time Series Analysis and Pattern Recognition
- AI in Experiments: Present Status and Future Prospects
- The Specialization of Intelligence in AI Horizons: Present Status and Visions for the Next Era
The research program is carried out partly within the framework of the NATIONAL CENTRE FOR HPC, BIG DATA AND QUANTUM COMPUTING.
Contact person
Antonio Pagliaro
Team
Anna Anzalone
A. Alessio Compagnino
Giancarlo Cusumano
Antonino La Barbera
Valentina La Parola
Antonio Pagliaro
Pierluca Sangiorgi
Antonio Tutone
Keywords
Artificial Intelligence · Machine Learning · Deep Learning · Computer Vision · Explainable AI · High Energy Astrophysics · Cherenkov Telescopes · ASTRI Mini-Array · Quantitative Finance · Time Series · Pattern Recognition