Gradient descent-based programming of analog in-memory computing cores
- 2022
- IEDM 2022
Bio
Julian Büchel is a Pre-Doc in the in-memory computing group at IBM Research Zurich. Büchel studied Computer Science (BSc) and Neural Systems and Computation (MSc) at ETH Zürich. During his MSc, Büchel worked at SynSense, a company focused on digital neuromorphic vision. Büchel joined IBM Research for his MSc thesis in 2021 and is a Pre-Doc since 2022. He is currently doing research in algorithms for analog in-memory computing, hardware-aware training methodologies, adversarial robustness/attacks, and generalization methods in deep learning.
Publications
NETWORK INSENSITIVITY TO PARAMETER NOISE VIA ADVERSARIAL REGULARIZATION
- Julian Büchel
- Fynn Faber
- et al.
- 2022
- ICLR 2022
ML-HW Co-Design of Noise-Robust TinyML Models and Always-On Analog Compute-in-Memory Edge Accelerator
- Chuteng Zhou
- Fernando Garcia Redondo
- et al.
- 2021
- IEEE Micro
Projects
Top collaborators
AS
Abu Sebastian
Distinguished Scientist & Manager
ML
Manuel Le Gallo-Bourdeau
Researcher in Neuromorphic and In-Memory Computing
VN
Vijay Narayanan
IBM Fellow & Strategist, Physics of AI. Senior Manager, PCM and AI Materials
AV
