Titre:
Accelerating Deep Learning with Approximate and Analog In-Memory Computing Principals
Conférencier:
Kaoutar El Maghraoui ,
IBM T.J Watson Research Center
Lieu:
Polytechnique Montréal, Pavillon J.-Armand Bombardier, local 1035 ,
Date et heure:
vendredi le 26 mai 2023 de
10:30 à 12:00
Résumé: The large investments in AI have led to a super-linear growth in data, models, and infrastructure capacity. We are also at an inflection point with the emergence of large scale generalizable and adaptable foundation models. This self-accelerated growth in AI is leading to increased size, and complexity of AI models, exa-scale computational demands and much increased carbon footprint. Hardware specialization and acceleration are key to satiate the computational demands of DNNs, which require synergistic cross-layer design across different layers of the compute stack. This talk describes a holistic approach to designing specialized AI systems pioneered by IBM Research. It highlights a suite of techniques towards the design & build of efficient deep learning systems. This involves approximate computing principals and Analog In-Memory non-Von-Neumann approaches to unlock exponential gains of AI computations making AI faster, more efficient, and sustainable.
Note biographique: Dr. Kaoutar El Maghraoui is a principal research scientist at the IBM T.J Watson Research Center and Adjunct Professor of Computer Science at Columbia University. Kaoutar’s work lies at the intersection of systems and artificial intelligence (AI). She leads the AI testbed of IBM Research AI Hardware Center, a global research hub focusing on enabling efficient next-generation accelerators and systems for AI workloads. She is responsible for the open-source development and cloud end user experience for IBM’s emerging Digital and Analog AI accelerators. Kaoutar has co-authored several patents, conference, and journal publications in the areas of systems research, distributed systems, high performance computing, and AI. She holds a PhD. degree from Rensselaer Polytechnic Institute, USA, and a bachelor’s and master’s Degrees from Al Akhawayn University, Morocco. She was recognized with several awards including the Robert McNaughton Award for best thesis in computer science, Best of IBM award in 2021, IBM’s Eminence and Excellence award for leadership in increasing Women’s presence in science and technology, several IBM outstanding technical accomplishments, 2021 IEEE TCSVC Women in Service Computing award, and 2022 IBM Technical Corporate award. Kaoutar is the global vice-chair of the Arab Women in Computing organization and active member in many women in science and technology initiatives. She is an ACM Distinguished Member, Senior IEEE Member, and member of Society of Women Engineers, TinyML foundation, and AnitaB.org.