About Me

Dal Pozzolo Andrea

I am a data scientist focusing on machine learning for fraud detection.

My Career

Senior Enterprise Data Scientist

At PMI I work on the identification of consumption patterns of Reduced Risk Products. The goal is to understand how people use smoke-free products to help converting a higher number of customers to less harmfull alternatives than standard cigarettes.

Jan. 2017

Philip Morris International, Lausanne, Switzerland

Senior Consultant in the Quant & Analytics team

At Nordea Bank I used Spark ML to prioritize money laundering alerts and reduce investigation cost. Prior to that I worked at UBS were I used Supervised and Unsupervised Machine Learning techniques to detection Money Laundering activities. I also audited the credit risk model for VaR and Stressed VaR of UBS. In a different context, I applied clustering-based anomaly detection algoirithms to identify abnormal Balance Sheets and Income statements collected by Global Format GmbH. At Credit Suisse I worked on the identification of fraudulent External Assets Managers and at Zurich Insurance I worked on identification of treasury fraud patterns using Unsupervised Learning algorithms.

Apr. 2016 - Dec. 2017

Ernst & Young, Zurich, Switzerland

Decision Analytics Consultant

At VASCO I worked on the development of a Link Analysis application for fraud investigations and supported the design of a Fraud Detection systems based on Machine Learning algorithms.

Jan. 2016 - Mar. 2016

VASCO Data Security, Wemmel, Belgium

PhD student in Machine Learning

PhD student at the Machine Learning Group (MLG) of the Université Libre de Bruxelles (ULB) under the supervision of Prof. Gianluca Bontempi. Collaboration with Prof. Nitesh Chawla (University of Notre Dame, Indiana, USA) and with Prof. Giacomo Boracchi (Politecnico di Milano, Milan, Italy).

Jan. 2012 - Dec. 2015

Machine Learning Group - ULB, Brussels, Belgium

Analyst

Supervision of VisualTrader development (online stock trading platform). Customer support of the trading platform. Development of trading systems and market scanners.

Aug. 2009 - Jul. 2011

Directa Sim, Turin, Italy

Master student in Statistics

In Decembre 2011, I graduated from a Master of Science (M.S.) in Applied Statistics and Actuarial Science at Università di Bologna with 110/110 Summa Cum Laude. Master thesis written at the Machine Learning Group of the Université Libre de Bruxelles.

Jan. 2009 - Dec. 2011

Master in Applied Statistics and Actuarial Science - UniBO, Bologna, Italy

Actuary Assistant

Balance analysis and reporting according to US GAAP and IFRS regulation (HB2).

Nov. 2008 - Feb. 2009

Allianz, Milan, Italy

Bachelor student in Statistics

In July 2009, I graduated from a Bachelor in Statistics for Finance and Insurance at Università di Bologna with 110/110 Summa Cum Laude. Second year at University of Glasgow (UK) as an Erasmus student.

Sep. 2006 - Jul. 2009

Bachelor in Statistics for Finance and Insurance - UniBO, Bologna, Italy

My Skills

Research

Technical

Machine Learning

Data Mining


Fraud Detection

R


Python

Big Data


Apache Spark

Apache Hadoop


SQL

C/C++


Java

Git


Latex

HTML/CSS


My Projects

unbalanced: Racing for Unbalanced Methods Selection

R package to tackle the problem of unbalanced classification by performing under- and over-sampling with various techniques.

Adaptive Machine Learning for Credit Card Fraud Detection

Resources on the topic of Machine Learning for Credit Card Fraud Detection.


My Publications

2017

  • Credit Card Fraud Detection: a Realistic Modeling and a Novel Learning Strategy

    A. Dal Pozzolo, G. Boracchi, O. Caelen, C. Alippi and G. Bontempi
    IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2017. [pdf]
  • SCARFF: a Scalable Framework for Streaming Credit Card Fraud Detection with Spark

    F. Carcillo, A. Dal Pozzolo, Y. Le Borgne, O. Caelen, Y. Mazzer and G. Bontempi
    Information Fusion, Elsevier, 2017. [pdf]

2015

  • When is undersampling effective in unbalanced classification tasks?

    A. Dal Pozzolo, O. Caelen, and G. Bontempi
    Machine Learning and Knowledge Discovery in Databases (ECML-PKDD), Porto, Portugal, 2015 [Paper] [Slides] [Poster]
  • Calibrating Probability with Undersampling for Unbalanced Classification

    A. Dal Pozzolo, O. Caelen, R. A Johnson and G. Bontempi
    IEEE Symposium Series on Computational Intelligence (SSCI), Cape Town, South Africa, 2015. [Paper] [Slides]
  • Credit Card Fraud Detection and Concept-Drift Adaptation with Delayed Supervised Information

    A. Dal Pozzolo, G. Boracchi, O. Caelen, C. Alippi and G. Bontempi
    International Joint Conference on Neural Networks (IJCNN), Killarney, Ireland, 2015. [Paper] [Slides]

2014

  • Learned lessons in credit card fraud detection from a practitioner perspective

    A. Dal Pozzolo, O. Caelen, Y. Le Borgne, S. Waterschoot, and G. Bontempi
    Expert Systems with Applications, vol. 41, no. 10, pp. 4915–4928, 2014. [pdf]
  • Using HDDT to avoid instances propagation in unbalanced and evolving data streams

    A. Dal Pozzolo, R. A Johnson, O. Caelen, S. Waterschoot, N. V Chawla, and G. Bontempi
    International Joint Conference on Neural Networks (IJCNN), Beijing, China, 2014. [Paper] [Slides]

2013

  • Racing for unbalanced methods selection

    A. Dal Pozzolo, O. Caelen, S. Waterschoot, and G. Bontempi
    International Conference on Intelligent Data Engineering and Automated Learning (IDEAL), Hefei, China, 2013. [Paper] [Slides] [Rpackage]

Thesis

  • Adaptive Machine Learning for Credit Card Fraud Detection

    A. Dal Pozzolo, and G. Bontempi
    PhD in Computer Science at Universite Libre de Bruxelles, 2015 [pdf] [Slides]
  • Comparison of Data Mining Techniques for Insurance Claim Prediction

    A. Dal Pozzolo, G. Moro, and G. Bontempi
    Master of Science in Applied Statistics and Actuarial Science at Universita di Bologna, 2011 [pdf]

© 2019 Andrea Dal Pozzolo