Autonomous Machine Learning


Making perfomant and robust machine learning models without (much) human intervention. One of my first approaches on the subject was my PhD thesis where I tried to evolve reservoir topologies (Echo State Networks) to better fit the task at hand.

Related Publications


conference Eleni Nisioti, Kyriakos C. Chatzidimitriou, Andreas L. Symeonidis: Predicting hyperparameters from meta-features in binary classification problems, 2018, AutoML 2018: International Workshop on Automatic Machine Learning, collocated with the Federated AI Meeting (ICML, IJCAI, AMAS, and ICCBR) [PDF] [Code]

Acceptance rate: 39 out of 65 submissions (=60%)


journal Kyriakos C. Chatzidimitriou, Pericles A. Mitkas: Adaptive reservoir computing through evolution and learning, 2013, Neurocomputing, 103, pp. 198-209 [Code] [DOI: 10.1016/j.neucom.2012.09.022]


thesis Kyriakos C. Chatzidimitriou: Reinforcement learning and evolutionary computing mechanisms for autonomous agents, 2012 [PDF]

My PhD dissertation (in Greek)


conference Kyriakos C. Chatzidimitriou, Anthony C. Chrysopoulos, Andreas L. Symeonidis, Pericles A. Mitkas: Enhancing Agent Intelligence through Evolving Reservoir Networks for Predictions in Power Stock Markets, 2011, ADMI 2011, pp. 228-247 [DOI: 10.1007/978-3-642-27609-5_15]
conference Kyriakos C. Chatzidimitriou, Ioannis Partalas, Pericles A. Mitkas, Ioannis P. Vlahavas: Transferring Evolved Reservoir Features in Reinforcement Learning Tasks, 2011, European Workshop on Reinforcement Learning (EWRL) 2011, pp. 213-224 [DOI: 10.1007/978-3-642-29946-9_22]


other Kyriakos C. Chatzidimitriou, Fotis E. Psomopoulos, Pericles A. Mitkas: Grid-enabled parameter initialization for high performance machine learning tasks, 2010, 5th EGEE User Forum [PDF]
conference Kyriakos C. Chatzidimitriou, Pericles A. Mitkas: A NEAT Way for Evolving Echo State Networks, 2010, European Conference on Artificial Intelligence (ECAI) 2010, pp. 909-914 [DOI: 10.3233/978-1-60750-606-5-909]

Diploma Theses

I've worked on the subject with diploma theses students in the following proejcts:

  • ADS: Automated data scientist (AutoML) in R for binary classification by Eleni Nisioti. [GitHub]