BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:EPFL SIAM Student Chapter Seminar on AlphaTensor -  Francisco Ruiz
  (Google DeepMind)
DTSTART:20231205T180000
DTEND:20231205T190000
DTSTAMP:20260512T015106Z
UID:e0b6f71b95029717d299a935dbad9733bfa9f01bdfab7cd86eeb0621
CATEGORIES:Conferences - Seminars
DESCRIPTION:Francisco Ruiz\, Research Scientist @ Google DeepMind London
  (UK)\nThe EPFL SIAM Student Chapter is honored to host Francisco Ruiz\, 
 Research Scientist @ Google DeepMind London (UK)\, who will give an onli
 ne seminar on AlphaTensor. The event will be projected in MA B1 11.\n\nFra
 ncisco Ruiz is a Research Scientist working at Google DeepMind in the Deep
  Learning Team. Before joining DeepMind\, he was a Postdoctoral Research S
 cientist in the Department of Computer Science at Columbia University and 
 in the Engineering Department at the University of Cambridge\, where he he
 ld a Marie-Sklodowska Curie fellowship in the context of the E.U. Horizon 
 2020 program. He completed his Ph.D. and M.Sc. from the University Carlos 
 III in Madrid. His research is focused on statistical machine learning\, e
 specially Bayesian modeling and inference.\n\n\n\n\nTitle of the Seminar:
  AlphaTensor: Discovering faster matrix multiplication algorithms with re
 inforcement learning\n\nAbstract: Algorithms are the building blocks of c
 omputation\, and their efficiency in solving fundamental computational tas
 ks is crucial. Matrix multiplication is one such primitive task\, occurrin
 g in many systems - from machine learning to scientific computing routines
 . Despite decades of research\, the question of how efficiently we can mul
 tiply n x m by m x p matrices remains open. Here we develop a Deep Reinfor
 cement Learning approach for discovering efficient and provably correct ma
 trix multiplication algorithms. Our agent\, AlphaTensor\, is trained to pl
 ay a single-player tensor decomposition game with a finite factor space\, 
 and yields algorithms for multiplying arbitrary matrices. AlphaTensor disc
 overed new algorithms that outperform the state-of-the-art complexity for 
 many matrix sizes (n\, m\, p). In particular\, AlphaTensor improved on Str
 assen's two-level algorithm for 4 x 4 matrices in a finite field\, which r
 epresents the first improvement since its discovery 50 years ago. We furth
 er demonstrate the flexibility of AlphaTensor through different use-cases:
  it discovered novel algorithms with state-of-the-art complexity for struc
 tured matrix multiplication\, and improved the practical efficiency of mat
 rix multiplication by optimizing for runtime on specific hardware. Our res
 ults highlight AlphaTensor's ability to accelerate the process of algorith
 mic discovery on a range of problems\, and optimizing for different criter
 ia.\n\nHope to see all of you there!\n\nIf you want to be informed about t
 he latest event of the EPFL SIAM Student Chapter\, please visit our websit
 e https://siam.epfl.ch or follow our LinkedIn page https://www.linkedin
 .com/company/epfl-siam-student-chapter/\nIf you want to join our associati
 on\, drop a message to siam@epfl.ch !
LOCATION:MA B1 11 https://plan.epfl.ch/?room==MA%20B1%2011 https://epfl.zo
 om.us/j/63049702950?pwd=eXJVbEdlZkJuLzlpc2h5MFpURVl5UT09
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR