The IceCube ML project

"The art of reconstruction - turning signal in a detector into an accurate estimate of what caused the signal - is a complex optimisation problem, for which Machine Learning and in particular Graph Neural Networks may be perfectly suited, if the processes behind the signal are understood from first principles and simulated accordingly."
[A possible mindset of the IceCube ML project]

Various performance plots (links to more below):

Group Members:

Troels C. Petersen Project part: Inspiration, physics, detector, and coordination. Period: First thoughts (with Andreas) in 2018. Type: Regular job! Goal: A great ML reconstruction, and fun getting there!		Andreas Soegaard Project part: Eventually, probably all parts Period: September 2021 (Marie-Curie Fellow). Email: andreas.soegaardnbi.ku.dk? Result: GitHub repository.
Morten Holm Project part: GNN reconstruction, Moon pointing analysis Period: February 2022 - December 2022 (Master Thesis). Email: qgf305alumni.ku.dk Result: GitHub repository.	Aske Luja Rosted Project part: GNN high energy neutrino reconstruction Period: October 2022 - October 2025 (Ph.D. at Chiba Univ.). Email: askerostedgmail.com Result: GitHub repository.
Peter Andresen Project part: GNN Moon pointing analysis Period: August 2022 - May 2023 (Master Thesis). Email: pqb128alumni.ku.dk Result: GitHub repository.		Andreas Mosgaard Joergensen Project part: GNN classification and reconstruction of high energy neutrinos Period: August 2022 - May 2023 (Master Thesis). Email: andreas.mosgaard.97@gmail.com Result: GitHub repository.
Simon Debes Project part: GNN classification and reconstruction of muons, MC-data calibration Period: August 2022 - May 2023 (Master Thesis). Email: rlq306alumni.ku.dk Result: GitHub repository.		Moust Holmes Project part: GNN classification and reconstruction of upgrade data Period: August 2022 - May 2023 (Master Thesis). Email: moust.holmes@gmail.com Result: GitHub repository.
Kaare Endrup Iversen Project part: GNN Upgrade reconstruction, Neutrino oscillation analysis Period: August 2021 - May 2022 (Master Thesis). Email: nvc889alumni.ku.dk Result: Master Thesis, Thesis Defence, and GitHub repository.		Leon Bozianu Project part: GNN classification and reconstruction of muons, MC-data calibration Period: August 2021 - May 2022 (Master Thesis). Email: qzr746@alumni.ku.dk Result: Master Thesis, Thesis Defence (Google Slides), Notes on files, and GitHub repository.

For an introduction talk to the GraphNet paper, one can consider a seminar talk given by Troels in the Oxford physics group on 25th of October 2022.
A very useful document is Etienne Bourbeau's Ph.D. thesis (2020), which gives a great introduction to both IceCube and data formats/level/etc. in general and neutrino oscillations in particular.
A great "primer" on neutrino oscillations in IceCube is Slides from Jason Koskinen (Summer School 2021), which gives a nice overview of the neutrino oscillation measurement in IceCube.

Previous group members:

Rasmus F. Oersoe
Project part: Graph Neural Net (PyTorch) reconstruction, data curration, etc.
Period: July 2020 - May 2021 (Master Thesis).
Email: pcs557alumni.ku.dk
Result: Master Thesis, Thesis Defence (Google Slides), and GitHub repository.
Haider Al-Saadi
Project part: Noise suppresion, (S)TGCN reconstruction?
Period: February - November 2021 (Master Thesis).
Email: krg825alumni.ku.dk
Result: GitHub repository.

Johann Bock Severin
Project part: GNN reconstruction (Keras TF), Angular Loss functions
Period: February - June 2021 (Bachelor Thesis)
Email: johann.bock.severingmail.com
Result: GitHub repository.
Christian Jespersen
Project part: GNN reconstruction (Keras TF)
Period: February - June 2021 (Bachelor Thesis).
Email: htd809alumni.ku.dk
Result: GitHub repository.

Jakob Schauser
Project part: GNN reconstruction (Keras TF)
Period: February - June 2021 (Bachelor Thesis).
Email: jakob.schausergmail.com
Result: GitHub repository.
Jonas Vinther
Project part: GNN reconstruction (PyTorch), Moon shadow analysis
Period: February - June 2021 (Bachelor Thesis).
Email: jv97live.dk
Result: GitHub repository.

Frederik V. S. S. Hansen
Project part: TCN reconstruction (Keras TF), Stopped Muon PID and reconstruction
Period: February - June 2021 (Bachelor Thesis).
Email: cjb924alumni.ku.dk
Result: GitHub repository.
Simon C. Debes
Project part: TCN reconstruction (Keras TF), Stopped Muon PID and reconstruction
Period: February - June 2021 (Bachelor Thesis)
Email: rlq306alumni.ku.dk
Result: GitHub repository.

Mads Ehrhorn
Project part: CNN and TCN reconstruction, data curration, etc.
Period: September 2019 - February 2021.
Results: Master Thesis, Thesis Defence, and GitHub repository. Bjoern Moelvig
Project part: RNN/GRU reconstruction, loss function exploration
Period: September 2019 - October 2020.
Results: Master Thesis, Thesis Defence, and GitHub repository.

Last updated 25th of September 2022.

	Rasmus F. Oersoe Project part: Graph Neural Net (PyTorch) reconstruction, data curration, etc. Period: July 2020 - May 2021 (Master Thesis). Email: pcs557alumni.ku.dk Result: Master Thesis, Thesis Defence (Google Slides), and GitHub repository.			Haider Al-Saadi Project part: Noise suppresion, (S)TGCN reconstruction? Period: February - November 2021 (Master Thesis). Email: krg825alumni.ku.dk Result: GitHub repository.
	Johann Bock Severin Project part: GNN reconstruction (Keras TF), Angular Loss functions Period: February - June 2021 (Bachelor Thesis) Email: johann.bock.severingmail.com Result: GitHub repository.			Christian Jespersen Project part: GNN reconstruction (Keras TF) Period: February - June 2021 (Bachelor Thesis). Email: htd809alumni.ku.dk Result: GitHub repository.
	Jakob Schauser Project part: GNN reconstruction (Keras TF) Period: February - June 2021 (Bachelor Thesis). Email: jakob.schausergmail.com Result: GitHub repository.			Jonas Vinther Project part: GNN reconstruction (PyTorch), Moon shadow analysis Period: February - June 2021 (Bachelor Thesis). Email: jv97live.dk Result: GitHub repository.
	Frederik V. S. S. Hansen Project part: TCN reconstruction (Keras TF), Stopped Muon PID and reconstruction Period: February - June 2021 (Bachelor Thesis). Email: cjb924alumni.ku.dk Result: GitHub repository.			Simon C. Debes Project part: TCN reconstruction (Keras TF), Stopped Muon PID and reconstruction Period: February - June 2021 (Bachelor Thesis) Email: rlq306alumni.ku.dk Result: GitHub repository.
	Mads Ehrhorn Project part: CNN and TCN reconstruction, data curration, etc. Period: September 2019 - February 2021. Results: Master Thesis, Thesis Defence, and GitHub repository.			Bjoern Moelvig Project part: RNN/GRU reconstruction, loss function exploration Period: September 2019 - October 2020. Results: Master Thesis, Thesis Defence, and GitHub repository.