University of Oxford, OxMLSys

I am working on the topic of Deep Learning efficiency in the context of embedded systems with Dr. Nic Lane. I specialise on Compression and Neural Architecture Search for mobile and MCU platforms. My current working papers are:

Svoboda F., Lane N., Focused Constrained Neural Architecture Search (2020)

Svoboda F., Liberis E., Lane N., In search of theoretically founded pruning. (2019)

Svoboda F. et al, Resource Efficient Deep Reinforcement Learning
for Acutely Constrained Mobile and Embedded Devices (2018)

The foundation of this work was laid at UCL during my MSc in Computational Statistics and Machine Learning, under the supervision of the same Dr. Nic Lane.

University of Oxford, OMI

In my graduate studies, I worked under the supervision of Dr. James Wolter on the development of objective-aware model selection strategies for the portfolio selection pipeline. It remains un-published as it is propietary.

University of Amsterdam

In collaboration with my undergraduate supervisor Dr. Tomislav Ladika I worked on the analysis of syndicated loan renegotiation in the US market. The dataset that I complied formed the basis of his further work. Most notably the:

Campello M., Ladika T. Matta R., Renegotiation Frictions and Financial Distress Resolution: Evidence from CDS Spreads (2015)

Early work

During my high school I participated in an Oncology research project at the Experimental Oncology Institute of the Slovak Academy of Sciences, for which I was awarded:

  • first place in Young Oncologist Competition (Slovakia)
  • second place in Slovak National Biology Olympics
  • Bronze medal in INEPO Eurasia (Azerbaijan)

In addition, The Slovak Minister of Education awarded me the Memorial List of St. Gorazd, the highest student academic state award in Slovakia.
My work was published as:

Svoboda F., Antioxidant effects of Fisetin, Young Oncologist Competition Almanac 6.-7. of March 2012 (2012)