Mehmet Emre Andıran

Research interests

Signal processing and machine learning

Both the theory and its applications, across imaging, vision, networks, time series, and sensing and communication systems.

Mathematics

Especially probability, statistics, and game theory.

Publications

1. ExpandDiff: Dynamic Range Expanding Diffusion for Single-Image HDR Reconstruction

   M. E. Andıran, Z. Yang, L. Lu, M. Salzmann, S. Süsstrunk

   Submitted to IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2027  Under review

2. Time Series Classification via Enriched Visibility Graph Feature Extraction

   E. Y. Pişkin, M. E. Andiran, E. Kulkul, A. Koç

   34th IEEE Conference on Signal Processing and Communications Applications (SIU), Istanbul, Turkey, 2026

   • DOI

Education

• B.S. Electrical and Electronics Engineering

  Bilkent University, Ankara, Turkey Sep 2023 – Jun 2027 (expected)

  • GPA 3.99 / 4.00, ranked 2nd in a cohort of 252. High Honour student every semester.
  • Full merit scholarship for undergraduate studies.

• Exchange Student, Electrical and Electronics Engineering

  EPFL, Lausanne, Switzerland Feb – Jul 2026

  • GPA 5.83 / 6.00. Swiss-European Mobility Programme (SEMP) scholarship recipient.
  • Selected coursework: Computer Vision, Reinforcement Learning, Network Machine Learning.

Experience

• Undergraduate Researcher

  Image and Visual Representation Lab (IVRL), EPFL, Lausanne, Switzerland Feb 2026 – Present

  • Developed ExpandDiff, a conditional diffusion model that jointly reconstructs clipped shadows and highlights from a single LDR image (first author, ICASSP 2027 submission), supervised by Zhuoqian Yang and Liying Lu in the lab led by Prof. Sabine Süsstrunk.
  • Introduced randomized-clipping training, so one model handles inputs with varying and unknown clipping levels, and trained it to predict HDR in a perceptually uniform encoding through a bounded output head.
  • Outperforms the strongest evaluated competing method in PU21-PSNR by 3.43 dB on the SI-HDR benchmark, and by 7.34 dB when shadows and highlights are both clipped.

• Undergraduate Researcher

  Koç Lab, Bilkent University, Ankara, Turkey Sep 2025 – Present

  • Writing a first-author paper on spatio-temporal forecasting over sensor networks with graph signal processing, supervised by Assoc. Prof. Aykut Koç.
  • Co-authored a paper on time series classification with visibility-graph features (IEEE SIU 2026).

• Summer Intern, Navigation Systems Design Engineering

  ASELSAN, Ankara, Turkey Jul – Aug 2026

  • Developed a causal temporal convolutional network (TCN) to estimate IMU sensor bias drift during and after transfer alignment.
  • Built an end-to-end simulation pipeline for a terrain-aided inertial navigation system, with a learning-based model that corrects position drift from radar altimeter data and terrain heightmaps.

• Summer Intern

  National Magnetic Resonance Research Center (UMRAM), Ankara, Turkey Jun – Jul 2025

  • Wrote an MLP framework from scratch and applied it to MNIST digit classification, reaching ~98% accuracy. Forward and backward propagation, gradient descent, regularization, and hyperparameter tuning were all implemented by hand.

• Teaching Assistant & Peer Tutor

  Bilkent University, Ankara, Turkey Sep 2024 – May 2025

  • Peer Tutor for MATH 101 (Calculus I), selected by the University Administration from a pool of candidates; led solo recitation sessions.
  • Teaching Assistant for CS 115 (Introduction to Programming in Python), supporting students through laboratory assignments.

• Vice President

  IEEE Bilkent University Student Branch, Ankara, Turkey Jun 2025 – Present

  • Lead a team of 25 active members organizing seminars, technical workshops, and summits in engineering and computer science for 300+ participants.
  • Previously Robotics & Automation Society Co-Chair (Aug 2024 – Jun 2025): taught a five-session Arduino workshop to 50+ attendees and designed PCBs for electronics workshops.

Projects

• Relation-Aware Graph Attention for NBA Trajectory Forecasting

  A group project for EPFL's Network Machine Learning course: a typed-edge graph attention network that forecasts player and ball trajectories.

  • PyTorch
  • Graph neural networks

• Inverse Design of Mie-Scattering Nanoparticles

  A group project for EPFL's Reinforcement Learning course, with the LNET lab: using reinforcement learning to design nanoparticles with target absorption and scattering cross-sections.

  • Reinforcement learning
  • Python

• 3-bit Analog-to-Digital Converter

  A 16 kHz sampling-rate ADC designed and built from discrete transistors, with a comparator, multiplexer, and sample-and-hold stage. An Arduino drives and monitors the system.

  • Discrete transistors
  • Arduino
  • LTspice

• Basketball Arcade Game

  An arcade cabinet on an ARM Cortex-M0 microcontroller: IR-sensor score keeping, a servo-driven moving hoop with an analog speed selector, buzzer music, and Bluetooth logging.

  • C
  • ARM Cortex-M0
  • Bluetooth

• Superheterodyne Analog Transceiver

  A 27 MHz transceiver with a 15 MHz intermediate frequency, built for an analog electronics course. By the end of the term it was good enough to hold real conversations over.

  • Analog RF
  • PCB design

• Password-Locked Safe with Intruder Alarm

  A safe implemented in VHDL on a BASYS3 FPGA, combining analog piezo sensing for intruder detection with a keypad for password entry and changes.

  • VHDL
  • BASYS3 FPGA
  • Vivado

Skills

Programming

Python, MATLAB, C, Assembly, VHDL, Arduino

Tools

PyTorch, MATLAB, LTspice, Vivado, KiCad, Proteus

Hardware

Microcontrollers, FPGAs, transistor-level design, PCBs, oscilloscopes, soldering

Languages

Turkish (native), English (C2, IELTS 8.5), German (elementary)