Gábor Balázs

gabalz@gandg.ai

I am a machine learning engineer and researcher focusing on regression and sequential decision-making. I currently work as a freelancer in Spain, hold the title of Competitions Expert on Kaggle, and pursue several self-funded scientific research projects.

I received my BSc and MSc at the Eötvös Lóránd University, and my PhD in Statistical Machine Learning at the University of Alberta under the supervision of Csaba Szepesvári and Dale Schuurmans. At the moment I live in Cartagena (Spain) with my wife Gema who is a writer and an illustrator.

Publications

• Adaptively partitioning max-affine estimators for convex regression,
  Gábor Balázs, AISTATS 2022, [paper] [5min talk]
  
  
• Convex regression: theory, practice, and applications,
  Gábor Balázs, PhD thesis, University of Alberta, 2016, [thesis]
  
  
• Near-optimal max-affine estimators for convex regression,
  Gábor Balázs, András György, Csaba Szepesvári, AISTATS 2015, [paper]

Preprints and Technical Reports

• Near-optimal delta-convex estimation of Lipschitz functions,
  Gábor Balázs, arXiv:2511.15615, 2025, [arXiv]
  
  
• Max-affine estimators for convex stochastic programming,
  Gábor Balázs, András György, Csaba Szepesvári, arXiv:1609.06331, 2016, [arXiv] [code]
  
  
• Chaining bounds for empirical risk minimization,
  Gábor Balázs, András György, Csaba Szepesvári, arXiv:1609.01872, 2016, [arXiv]
  
  
• Cascade-Correlation Neural Networks: A Survey,
  Gábor Balázs, UofA Technical Report 2009, [pdf]

Software

• Convex and Delta-Convex Regression (Python)
• Robot Segway Simulator (Java)
• RL-Glue Lisp Codec (Common Lisp)