Viktor Erdélyi

(Katakana: ヴィクトル エルデーイ)

Specially appointed assistant professor, Mobile Computing Laboratory (Yamaguchi Lab), Graduate School of Information Science and Technology, Osaka University, Japan

Contact: viktor [at] ist [dot] osaka-u [dot] ac [dot] jp

Introduction

Viktor Erdélyi received his BSc and MSc degrees in informatics from the Eotvos Lorand University, Hungary, and, in 2019, he received his doctoral degree (Dr.-Ing.) in informatics from the Max Planck Institute for Software Systems (MPI-SWS), Germany. Since 2019, he has been a specially appointed assistant professor at Mobile Computing Laboratory (Yamaguchi Lab) at the Graduate School of Information Science and Technology in Osaka University, Japan. Previously, he worked on research projects related to security and privacy, mobile computing, acoustic signal processing and localization, and he published his research results in top international conferences such as ACM MobiSys and USENIX Security. He is currently working on the following research topics: Smartphone-based detection of acute social withdrawal ("hikikomori") in Japanese university students Maintenance-free, passive sensing solutions for detecting pandemic-induced social isolation for elderly people in the US and Japan Human activity recognition using Wi-Fi CSI backscatter Localization of GHz-band backscatter tags Tag-free object recognition using microwave signals

Peer-reviewed scientific publications

Discussion series about the possibility of conscious computers

I had the honor to have some deep-dive discussions with Tom Campbell, physicist and author of the My Big TOE (MBT) trilogy. In this series, we discuss the concept of consciousness in the context of computers, how he views the possibility of conscious computers from the perspective of his MBT theory, how we could use recent advances in AI technology (such as deep learning, reinforcement learning, large language models) to build a conscious computer, and some speculations about what role conscious computers could play in a future society.

His webpage can be found at https://www.my-big-toe.com/ and http://www.tomcampbell.info/.

Previous projects

Sonoloc (PhD thesis project, 2014-2018)

This project is about scalable positioning of commodity mobile devices using audio signals. I developed a system that can compute a 2D relative position map of nearby mobile devices (such as smartphones). To compute the position map, the system uses a small number of acoustic signal transmissions emitted by a subset of the participating devices.

Source code

• Download link: Source code + documentation ZIP
• License: MIT license except for a few files written by third parties.

Internship at the National Institute of Informatics, Tokyo, Japan (2015)

A research internship under the supervision of Assoc. Prof. Nobutaka Ono. I implemented Nobutaka Ono's Harmonic-Percussive Sound Separation (HPSS) effect as an extension for Audacity (a popular open-source sound processing tool). The purpose of HPSS is to separate a music track into harmonic and percussive components (both of which are considered useful signals). I also implemented an HPSS-based vocal separation effect as a new effect for Audacity. The purpose of this effect is to separate vocal components from non-vocal components in a music track.

EnCore (2011-2014)

EnCore is a privacy-preserving platform for mobile social applications. It leverages rich contextual data available on modern mobile devices to provide social services in a privacy-preserving way.