Transit timing & ephemerides
We refine transit times and ephemerides from space- and ground-based observations, supporting reliable scheduling and long-baseline studies of planetary systems.
Research group · İstanbul, Türkiye
Exoplanet and Scientific Machine Learning
We study small signals in noisy scientific data: measuring exoplanet transits, identifying stellar-activity systematics, and building reliable machine-learning and Bayesian workflows that can be tested, reproduced, and reused.
QuestionsWhen do planets transit, and what can bias that measurement?
DataTESS, Kepler/K2, JWST and ground-based photometry
MethodsTransit models, Bayesian inference, Gaussian processes and ML
Research programme
Astronomy supplies the precision-measurement problems; scientific machine learning supplies tools for inference, diagnostics and scale. The group connects the two without treating either the data or the model as a black box.
Stellar activity & systematics
We model how starspots and instrumental effects distort measured transit times and depths, then propagate those uncertainties into the scientific result.
Scientific machine learning
We develop interpretable, validation-focused ML and Bayesian methods for time series, imaging and other scientific data where errors have physical meaning.
Atmospheres & small bodies
We contribute to exoplanet atmospheric-characterisation studies and multi-site stellar-occultation campaigns for trans-Neptunian objects.
Measure carefullySeparate weak physical signals from activity, cadence, noise and model assumptions.
Validate honestlyUse realistic tests, leakage controls and uncertainty estimates—not accuracy alone.
Work reproduciblyPrefer versioned code, traceable data products and results that can be re-derived.
Projects
Funded work is separated from proposals still under evaluation. Roles and funders are drawn from the principal investigator's maintained academic record.
Active funded work
- ML-based microscopic wood species identificationDeep-learning classification from microscopic imagery.
Completed funded work
Proposals under evaluation
People & participation
Research environment
Observational data
Space- and ground-based photometric time series, timing measurements and public scientific archives.
Computational methods
Python workflows for transit modelling, Bayesian analysis, Gaussian processes, simulation and machine learning.
Open practice
Version-controlled analysis, documented assumptions, reproducible notebooks and reusable teaching materials.
Selected research record
Generated from NASA ADS and filtered against ORCID and a maintained collision list. See the NASA ADS record for source metadata.
Teaching & research training
Current (2026–2027, İSTÜN): Physics I & II, Computer Programming I & II, Object-Oriented Programming, Data Structures & Algorithms, Robotics, and Machine Learning—in English and Turkish.
Student work can connect coursework to research through transit light curves, starspot simulations, scientific time-series modelling and reproducible Python pipelines.
Open materials are available through Algorithm Analysis with Python and the full course list.
Networks & scientific service
ExoClock Project
Ground-based observations and ephemeris refinement in support of ESA's Ariel mission; co-author contributions to ExoClock III and IV.
EXPLORE
Participation in international transit-timing work connecting new photometry with long-baseline exoplanet records.
Europlanet ML Working Group
Community work on machine learning in planetary science, including an invited contribution at EPSC-DPS 2025.
Multi-site campaigns
Analysis and coordination supporting physical characterisation of trans-Neptunian objects.
Education
Memberships
International Astronomical Union (IAU), European Astronomical Society (EAS), Turkish Astronomical Society (TAD), and European Association for Astronomy Education (EAAE). Referee for international astronomy and planetary-science journals.
Work with the group
Students at İSTÜN and researchers interested in exoplanet data analysis, transit timing, stellar activity or validation-focused machine learning are welcome to propose a focused project or collaboration.
Verified academic identity
The principal investigator's appointment is listed in the İSTÜN academic staff directory and Türkiye's YÖKSİS registry. Publications are linked through ORCID, NASA ADS and DOI records.
Short group description
The Exoplanet and Scientific Machine Learning Research Group, led by Arif Solmaz at İstanbul Health and Technology University, studies exoplanet transit timing, stellar-activity systematics and reliable computational inference. The group develops Bayesian and machine-learning workflows for TESS, Kepler/K2, JWST and ground-based observations, with an emphasis on validation, uncertainty and reproducible research.