PhD student at Lamarr Institute / TU Dortmund π§βπ Working on making ML & AI more trustworthy, resource-aware and sustainable π±π Musician on the side πΆ
Running some new experiments on ML energy consumption - comparing software-internal profiling (#CodeCarbon) with external power measuring!β‘ All in line to shed more light on #sustainableAI, as part of @LamarrInstitute's research on resource-aware ML! π
@strubell@JeffDean Quite shocking... If people would try to explicitly report their numbers on sustainability and resource usage, such (possibly faulty) estimates would not be necessary. Your work is important and valuable!
Well-arrived in #Vilnius for #ECMLPKDD! π Today presenting "MetaQuRe: #Metalearning from Model Quality and Resource Consumption" https://t.co/5JryNxYSGR and tomorrow "Towards More #Sustainable and #Trustworthy Reporting in Machine Learning" https://t.co/qnuvY9cQFI π¨βπ»π
Had lots of good conversations on #resourceaware ML and AI after yesterday's talk and at the poster session π±π Thank you for the valuable input and feedback, the full #openaccess paper is available here: https://t.co/nK3GOzQGBM
Hola #Barcelona - ready for #kdd2024@kdd_news, representing @LamarrInstitute π Tomorrow I will present our work on "Automated Multi-Objective Model Selection for Time Series Forecasting" - the #openaccess article is now also published! π¨βπ» https://t.co/nK3GOzQGBM
π£ Interviewpartner*innen gesucht π£
FΓΌr eine Studie zur "NΓΌtzlichkeit von #KI-Labeling" suchen wir Personen aus Industrie und Wirtschaft, die KI-Produkte entwickeln π¨βπ» Mehr Infos: QR-Code scannen oder https://t.co/9TEgNPut9V besuchen - bitte gerne teilen! π€
Machine Learning advances are only as reliable as their evaluations as the authors of the new #DAMI online-first paper stress: "Towards more sustainable and trustworthy reporting in machine learning" by Raphael Fischer, Thomas Liebig & @KatharinaJutta (https://t.co/3vxQGVdycN)#OA
With that the time of the #Lamarr team in #Pittsburgh concludes. π Today, our delegates are heading off to their final stops of the week in #StLouis, #Philadelphia and #NewYork. οΏ½οΏ½οΏ½ We are excited to see what great insights & collaborations they will return to Germany with. (3/3)
Final stop at @WUSTL is now over - three cities, five flights, conversations with about 100 people, all within seven days - but I enjoyed every second building international bridges with @LamarrInstitute π₯°
Pittsburgh you have been good to us! So much fruitful exchange with groups at @CarnegieMellon and @PittTweet, many opportunities for future research of our @LamarrInstitute. Today, onwards to St. Louis π¬
π Today, we we will follow up on these initial meetings and take up many more at @CarnegieMellon and the University of Pittsburgh (@PittTweet) where the #Lamarr delegates will engage in in-depth scientific working sessions with various research groups on-site. (3/3)
What a start into the week, kicking off our @LamarrInstitute US delegation visit (in collaboration with @uniallianz NYC)! π Great talks, lots of exchange, can't wait for the upcoming days in Pittsburgh and St. Louis. (HMU if you're in the area)!
Wondering how DL inference can be performed efficiently on the edge?π‘Our new paper unveils practical insights of utilizing USB accelerators by @intel and @GoogleAI https://t.co/gKaZ1LqVBr