Marcela Zuluaga
Postdoctoral Researcher
Computer Science Department
ETH Zurich

I am a postdoctoral researcher at ETH Zurich associated with the Advanced Computing Laboratory with the advise of Professor Markus PĆ¼schel

I have worked in two interesting projects during my stay here. The first one dealt with the automatic generation of hardware modules from high-level algorithmic representations. The aim is to leverage domain-specific knowledge to identify the parallelism and regularity that can be exploited during the hardware implementation stage. The flexibility of a high-level representation can be used to explore the different design trade-offs that exist. A summary of this work can be found on the website of the Spiral project and more details in the DAC paper listed under publications. A journal paper is under review.

The second project tackles the problem of predicting optimal solutions in a multi-objective design space. Here we developed an active learning algorithm that makes use of Gaussian process modelling to offer probabilistic guarantees on the efficiency and convergency of the algorithm. A summary of this work can be found on the website of the Spiral project and more details in the ICML paper listed under publications. A new version of the algorithm is soon to be submitted for review to the Journal of Machine Learning Research.

My main interest at the moment is applying machine learning and statistical methods to solve hard problems. I am looking now for opportunities to pursue this interest.