BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:CIDeR-ML General Meeting
DTSTART:20260508T000000Z
DTEND:20260509T024500Z
DTSTAMP:20260624T163200Z
UID:indico-event-10553@indico.slac.stanford.edu
DESCRIPTION:https://u-tokyo-ac-jp.zoom.us/j/83932834349\nLink to Recording
 \nMinutes:\nQuick recap\nThis meeting was a team check-in following a work
 shop and holiday period\, where team members provided updates on their rec
 ent work. Titouan presented improvements to a SIREN network model for PMT 
 acceptance correction\, sharing results of model performance optimization 
 including adjustments to omega values and addressing normalization issues\
 , with the model now showing 5% maximum error. To Titouan reported success
  in reducing computation time to approximately one millisecond per iterati
 on while working on lowering training losses\, though validation loss rema
 ined around 0.01 with some overfitting concerns. Other team members includ
 ing Zhenxiong\, Junjie\, and Ryotaro reported no updates\, while Yifan men
 tioned ongoing work on position-related tasks and basic checks.\nNext step
 s\n\nNahuel: send a message to Omar to confirm if the current one millisec
 ond per iteration step time is acceptable for the Adam optimizer step\nZhe
 nxiong: commit documented work and relevant notebooks/scripts to the repos
 itory\n\nSummary\nTeam Progress\nPatrick opened the meeting by welcoming p
 articipants after the holiday and workshop\, asking for reports from team 
 members. Zhenxiong reported having no updates this week. Patrick suggested
  documenting and committing the work done since the last workshop to a rep
 ository\, to which Zhenxiong confirmed she will do. Junjie had no addition
 al report to share.\nSIREN Neural Network Performance Issues\nTitouan disc
 ussed his work on training a SIREN neural network to learn corrections for
  PMT acceptance based on incident angle and hit position. He mentioned hav
 ing performance issues with his models and was investigating what might be
  causing them during the Golden Week. Adjusting omega helped with overfitt
 ing. He resolved normalization issues in the visualization and achieved a 
 maximum relative error of around 5%\, training in full 3D. Will continue t
 o investigate detailed structures in 2D error plots.\nNahuel reported on i
 mproving computation time to about one millisecond per iteration while wor
 king on the SIREN\, though the model showed some overfitting with validati
 on loss not decreasing below 0.01. He will investigate some potential reas
 ons for the overfitting\, such as still too complex central region\, issue
  from random sampling to line sample (try random sampling with new true fu
 nction)\, etc. The team agreed to check with Omar regarding the one-millis
 econd requirement for iterations and discussed the validation set approach
  used by Nahuel. \n\nhttps://indico.slac.stanford.edu/event/10553/
URL:https://indico.slac.stanford.edu/event/10553/
END:VEVENT
END:VCALENDAR