BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:CD Technical meeting (HD2): From Diagnostics to Deployment: Toward
 s Open\, FAIR Data and Software Infrastructure for AI-Driven Fusion Scienc
 e at UKAEA
DTSTART:20261104T150000Z
DTEND:20261104T160000Z
DTSTAMP:20261122T160400Z
UID:indico-event-866@indico.ukaea.uk
DESCRIPTION:Speakers: Samuel (Computing) Jackson (UKAEA)\n\nAbstract\nReal
 izing the promise of AI/ML for fusion energy requires infrastructure spann
 ing the full data lifecycle\, from raw diagnostic signals\, through curati
 on and benchmarking\, to real-time deployment in plasma control. This talk
  presents a progress on data\, infrastructure\, tools\, and models being d
 eveloped by UKAEA and collaborators to address this pipeline.\nWe start at
  the data layer\, where FAIR-MAST [7\, 6] and progress on its successor\, 
 the Fusion Data Service (FDS)\, provide open\, scalable access to fusion d
 iagnostic data and metadata data\, aligned with community standards such a
 s IMAS and DCAT\, and underpinned by a shared vocabulary for datasets\, de
 vices and workflows through the FUEL ontology [12]. For further interopera
 bility\, we developed libtokamap [13] to map experimental data into the IM
 AS schema.\nTurning raw signals into curated\, labelled\, AI-ready dataset
 s is the job of TokTagger [4]\, a open source\, web-based platform for int
 eractive labeling of diagnostic data of different modalities either by han
 d or from annotation models\, and NiceShot [5]\, an interactive dashboard 
 for exploring\, visualizing\, searching\, clustering\, and filtering relev
 ant shots by metadata across whole experimental campaigns.\nCurated data e
 nables downstream collaborations such as TokaMark[11]\, an open benchmark 
 of over 11\,000 real MAST shots across 14 different tasks\, including fore
 casting and reconstruction tasks across heterogeneous diagnostics covering
  magnetic\, kinetic\, radiative\, actuators\, and equilibrium data. From t
 his benchmark\, we built TokaMind [3]\, a multi-modal\, multi-fidelity tra
 nsformer model for fusion diagnostics that supports lightweight adaptation
  to new tasks.\nFinally\, we will turn to operational deployment through t
 he Real-Time Virtual Circuit (RTVC) programme [2\, 1\, 8\, 10\, 9]\, which
  replaces pre-computed lookup tables for tokamak shape control with neural
 -network emulators of the plasma response trained on synthetic FreeGSNKE e
 quilibria. RTVC preserves the interpretability and modularity of existing 
 control architectures while adapting in real time to the plasma state\, an
 d has reached experimental deployment on MAST Upgrade utilizing common emu
 lators across distinct scenarios without scenario-specific retraining.\nTo
 gether\, these projects trace a path from raw data to real-time AI in fusi
 on science. We will close by discussing the governance\, reproducibility a
 nd interoperability practices needed to sustain this infrastructure as it 
 scales across facilities and matures toward broader AI integration in fusi
 on workflows.\nReferences\n[1] A. Agnello et al. Emulation techniques for 
 scenario and classical control design of tokamak plasmas. Physics of Plasm
 as\, 31:043901\, 2024.\n[2] N. C. Amorisco et al. FreeGSNKE: A Python-base
 d dynamic free-boundary toroidal plasma equilibrium solver. Physics of Pla
 smas\, 31:042517\, 2024.\n[3] Tobia Boschi\, Rodrigo H. Ordonez-Hurtado\, 
 Eszter Sz´ekely\, Adriano Agnello\, Robert Akers\, Andrea Loreti\, C´eci
 le Rousseau\, Alexander Whittle\, Stanislas Pamela\, Juan Bernabe-Moreno\,
  Mykhaylo Zayats\, Nicola C. Amorisco\, George K. Holt\, Samuel Jackson\, 
 Alessandra Pascale\, and Vassil Alexandrov. TokaMind: A multi-modal transf
 ormer foundation model for tokamak plasma dynamics\, 2026.\n[4] S. Jackson
 \, M. Field\, J. Blake\, A. Saleem\, N. Bhatia\, R. Costa\, N. Cummings\, 
 S. Khan\, P. Sharma\, S. Pamela\, and A. Gonzalez-Beltran. TokTagger: An o
 pen source tool for labelling and tagging tokamak data\, 2026.\n[5] Samuel
  Jackson. NiceShot!: An interactive dashboard for exploring tokamak plasma
  shot data\, May 2026. Python package\; provides scatter plots\, correlati
 ons\, time traces\, filtering\, clustering\, and outlier detection over sh
 ot statistics.\n[6] Samuel Jackson\, Saiful Khan\, Nathan Cummings\, James
  Hodson\, Shaun de Witt\, Stanislas Pamela\, Rob Akers\, and Jeyan Thiyaga
 lingam. An open data service for supporting research in machine learning o
 n tokamak data. IEEE Transactions on Plasma Science\, 2025.\n[7] Samuel Ja
 ckson\, Saiful Khan\, Nathan Cummings\, James Hodson\, Shaun de Witt\, Sta
 nislas Pamela\, Rob Akers\, Jeyan Thiyagalingam\, and The MAST Team. FAIR-
 MAST: A fusion device data management system. SoftwareX\, 27:101869\, 2024
 .\n[8] K. Pentland et al. The FreeGSNKE pulse design tool (FPDT): a comput
 ational framework for evolutive plasma scenario and control design. Plasma
  Physics and Controlled Fusion\, 68:075035\, 2026.\n[9] K. Pentland et al.
  Real-time virtual circuits for plasma shape control via neural network em
 ulators: dynamic validation in closed-loop simulations\, 2026. Submitted.\
 n[10] A. Ross et al. Real-time virtual circuits for plasma shape control v
 ia neural network emulators\, 2026. Submitted.\n[11] Cecile Rousseau\, Sam
 uel Jackson\, Rodrigo H. Ordonez-Hurtado\, Nicola C. Amorisco\, Tobia Bosc
 hi\, George K. Holt\, Andrea Loreti\, Eszter Sz´ekely\, Alexander Whittle
 \, Adriano Agnello\, Stanislas Pamela\, Alessandra Pascale\, Robert Akers\
 , Juan Bernabe Moreno\, Sue Thorne\, and Mykhaylo Zayats. TokaMark: A comp
 rehensive benchmark for MAST tokamak plasma models\, 2026.\n[12] UKAEA. FU
 EL: FUsion Energy Lexicon\, an ontology for fusion energy data and process
 es\, 2026. Provides structured semantic definitions for fusion energy-rela
 ted information and workflows.\n[13] UKAEA. libtokamap: A C++20 library fo
 r flexible data mapping and transformation with support for multiple data 
 sources and mapping types\, 2026. Plugin-based architecture supporting JSO
 N-configured value\, data source\, expression\, dimension\, and custom map
 pings\, with built-in caching.\n\nhttps://indico.ukaea.uk/event/866/
URL:https://indico.ukaea.uk/event/866/
END:VEVENT
END:VCALENDAR
