Hamilton Institute logo Maynooth University logo

Atılım Güneş Baydin, PhD

Postdoctoral Researcher
Hamilton Institute & Department of Computer Science
National University of Ireland Maynooth
Maynooth, Co. Kildare, Ireland

E-mail: gunes@cs.nuim.ie
http://www.cs.nuim.ie/~gunes/

Atilim Gunes Baydin

Moved

ON 11 APRIL 2016, I JOINED THE MACHINE LEARNING RESEARCH GROUP AT THE UNIVERSITY OF OXFORD.
PLEASE VISIT THIS LINK.

About

I am a postdoctoral researcher at the Brain and Computation Lab, Hamilton Institute and the Department of Computer Science of the National University of Ireland Maynooth.

My current work, with Prof. Barak Pearlmutter, involves automatic differentiation (AD) and its applications in machine learning. Our research aims to add exact first-class differentiation operators to the lambda calculus, allowing numeric algorithms and scientific computations to be expressed in a very clear and succinct way. We are designing and implementing languages embodying compositionality, where gradient and hyperparameter optimization can be performed efficiently and concurrently on many nested levels in a system.

I received my PhD in artificial intelligence from Universitat Autònoma de Barcelona (UAB), in Barcelona, Spain. My research involved analogical and commonsense reasoning and also introduced a novel graph-based evolutionary algorithm employing semantic networks, paralleling sociological theories of evolutionary epistemology and memetics. During my PhD, I had been working under the supervision of Prof. Ramon Lopez de Mantaras at the Artificial Intelligence Research Institute (IIIA) of the Spanish National Research Council (CSIC) and I also had a chance to visit the Learning Research & Development Center of the University of Pittsburgh.

Before my work in Spain, I received my master's degree from the Complex Adaptive Systems program at the Department of Applied Physics of Chalmers University of Technology, in Göteborg, Sweden. Working under the Programmable Artificial Cell Evolution EU project, my work involved mesoscopic molecular modeling and computational physics.

Research Interests

Evolutionary computation, artificial life, complex systems, nonlinear dynamics, and neural networks.

Honors & Awards

Selected Talks

  • "Differentiable Programming" [pdf]
    Microsoft Research Cambridge, Cambridge, UK, February 1, 2016
  • "Automatic differentiation and machine learning"
    Department of Statistics, University of Oxford, Oxford, UK, March 9, 2015
  • "Automatic differentiation and machine learning"
    Microsoft Research Cambridge, Cambridge, UK, March 6, 2015

Recent / Upcoming Talks and Visits

Tools

I am the author of several tools. [GitHub profile]

Hype
An experimental library for deep learning and hyperparameter optimization.
Web site
GitHub

DiffSharp
A functional automatic differentiation library for machine learning.
Web site
GitHub

FsAlg
A generic linear algebra library.
Web site
GitHub

LambdaReactor
An implementation of the lambda calculus abstract chemistry of Walter Fontana and Leo Buss.
Website
GitHub

CPGEvolution
Evolution of central pattern generator (CPG) networks for bipedal walking.
YouTube video
GitHub

Publications

Atılım Güneş Baydin and Barak A. Pearlmutter.
Recurrent neural networks: origins, development, influence.
In prep.

Atılım Güneş Baydin, Barak A. Pearlmutter, and Jeffrey Mark Siskind.
DiffSharp: automatic differentiation library.
Submitted. [arXiv:1511.07727]

Atılım Güneş Baydin and Ramon López de Mántaras.
Evolutionary generative adaptation for case-based reasoning.
In prep.

Atılım Güneş Baydin, Barak A. Pearlmutter, Alexey Andreyevich Radul, and Jeffrey Mark Siskind.
Automatic differentiation in machine learning: a survey.
Submitted. [arXiv:1502.05767]

Atılım Güneş Baydin, Ramon López de Mántaras, and Santiago Ontañón.
A semantic network-based evolutionary algorithm for computational creativity.
Evolutionary Intelligence, 8(1):3–21, 2015. [doi:10.1007/s12065-014-0119-1] [arXiv:1404.7765]

Atılım Güneş Baydin and Barak A. Pearlmutter.
An analysis of publication venues for automatic differentiation research.
Internal report, 2014. [arXiv:1409.7316]

Atılım Güneş Baydin and Barak A. Pearlmutter.
Automatic differentiation of algorithms for machine learning.
In Proceedings of the AutoML Workshop at the International Conference on Machine Learning (ICML), Beijing, China, June 21–26, 2014. [arXiv:1404.7456] [web]

Atılım Güneş Baydin.
Evolutionary Adaptation in Case-Based Reasoning: An Application to Inter-Domain Analogies for Mediation.
PhD thesis, Institut d’Investigació en Intel·ligència Artificial, IIIA, Consejo Superior de Investigaciones Científicas, CSIC & Departament de Ciències de la Computació, Universitat Autònoma de Barcelona, Barcelona, Spain, 2013. [doi:10803/129294] [pdf]

Atılım Güneş Baydin, Ramon López de Mántaras, and Santiago Ontañón.
Automated generation of cross-domain analogies via evolutionary computation.
In Maher, M. L., Hammond, K., Pease, A., Pérez y Pérez, R., Ventura, D., and Wiggins, G., editors, Proceedings of the Third International Conference on Computational Creativity, Dublin, Ireland, May 30–June 1, 2012, pages 25–32. University College Dublin. [arXiv:1204.2335] [pdf]
(Best Student Paper Award presented by the Cognitive Science Society)

Atılım Güneş Baydin and Ramon López de Mántaras.
Evolution of ideas: A novel memetic algorithm based on semantic networks.
In Proceedings of the IEEE Congress on Evolutionary Computation, CEC 2012, IEEE World Congress On Computational Intelligence, WCCI 2012, Brisbane, Australia, June 10–15 2012, pages 2653–2660. IEEE Press. [arXiv:1201.2706] [doi:10.1109/CEC.2012.6252886]

Atılım Güneş Baydin.
Evolution of central pattern generators for the control of a five-link bipedal walking mechanism.
Paladyn Journal of Behavioral Robotics, 3(1):45–53, 2012. [arXiv:0801.0830] [doi:10.2478/s13230-012-0019-y]

Atılım Güneş Baydin, Ramon López de Mántaras, Simeon Simoff, Carles Sierra.
CBR with commonsense reasoning and structure mapping: An application to mediation.
In Ram, A. and Wiratunga, N., editors, Case-Based Reasoning Research and Development, volume 6880 of Lecture Notes in Artificial Intelligence, pages 378–392. Springer, Heidelberg, 2011. [arXiv:1108.0039] [doi:10.1007/978-3-642-23291-6_28]

Atılım Güneş Baydin.
Dissipative Particle Dynamics and Coarse-Graining: Review of Existing Techniques, Trials with Evolutionary Computation.
Master’s thesis, Department of Applied Physics, Chalmers University of Technology, Göteborg, Sweden, 2008. [pdf]

Melek Tendürüs, Atılım Güneş Baydin, Marieke A. Eleveld, Alison J. Gilbert.
City versus wetland: Predicting urban growth in the Vecht area with a simple cellular automaton model.
Submitted. [arXiv:1304.1609]

Connections

Last updated 27/04/2016