Florian Scheidegger is a Research Staff Member at the IBM Research laboratory in Zurich, researching and developing AI solutions for inspecting Civil infrastructures. Florian invented, developed, and disclosed core technologies that improve the detection performance of deep learning models for challenging client use-cases in that domain. With the help of the team and cross-laboratory collaboration he transfers and integrates the latest technologies into IBM products.
Florian Scheidegger graduated from ETH in 2020 with a PhD in Electrical Engineering and Information Technology. He has pursued an industrial PhD within IBM since 2017 working on transprecison computing contributing to the OPRECOMP project. He developed, implemented, and published reduced precision concepts that overcome over-conservative “precise” computing assumptions.
Florian Scheidegger achieved a Master of Science ETH in Electrical Engineering and Information Technology in 2017 with a main focus on software and hardware development for high-performance data processing applications. He worked for five months at the Integrated Systems Laboratory of ETH developing a spatio-temporal video pipeline in the first part, followed by a work that uses neural networks to classify videos.
- Constrained deep neural network architecture search for IoT devices accounting hardware calibration
Florian Scheidegger, Luca Benini, Costas Bekas, Cristiano Malossi
NeurIPS - Thirty-third Conference on Neural Information Processing Systems, 2019, 2019
(Acceptance rate 21.2% over 6743 reviewed submissions)
FloatX: A C++ Library for Customized Floating-Point ArithmeticACM Transactions on Mathematical Software (TOMS) 45(4), 1--23, ACM New York, NY, USA, 2019
Flegar, Goran, Florian Scheidegger, Vedran Novaković, Giovani Mariani, Andrés E. Tomas, A. Cristiano I. Malossi, and Enrique S. Quintana-Ortí