Timezone: »

Kernelized Synaptic Weight Matrices
Lorenz Müller · Julien Martel · Giacomo Indiveri

Thu Jul 12 09:15 AM -- 12:00 PM (PDT) @ Hall B #218

In this paper we introduce a novel neural network architecture, in which weight matrices are re-parametrized in terms of low-dimensional vectors, interacting through kernel functions. A layer of our network can be interpreted as introducing a (potentially infinitely wide) linear layer between input and output. We describe the theory underpinning this model and validate it with concrete examples, exploring how it can be used to impose structure on neural networks in diverse applications ranging from data visualization to recommender systems. We achieve state-of-the-art performance in a collaborative filtering task (MovieLens).

Author Information

Lorenz Müller (ETH Zurich and University of Zurich)
Julien Martel (ETH Zurich)

I am interested in vision systems that deport more intelligence on the sensors' focal plane. I have been working with sensors that collocate in their pixels some small circuitry next to the photosensitive element allowing them to sense more than just ``light intensity". I am interested in the development of algorithms and systems to exploit those.

Giacomo Indiveri (University of Zurich)

Giacomo Indiveri is a Professor at the Faculty of Science of the University of Zurich, Switzerland, and director of the Institute of Neuroinformatics of the University of Zurich and ETH Zurich. He obtained an M.Sc. degree in electrical engineering and a Ph.D. degree in computer science from the University of Genoa, Italy. He was a post-doctoral research fellow in the Division of Biology at Caltech and at the Institute of Neuroinformatics of the University of Zurich and ETH Zurich. He holds a "habilitation" in Neuromorphic Engineering at the ETH Zurich Department of Information Technology and Electrical Engineering. He was awarded an ERC Starting Grant on "Neuromorphic processors" in 2011 and an ERC Consolidator Grant on neuromorphic cognitive agents in 2016. His research interests lie in the study of neural computation, with a particular focus on spike-based learning and selective attention mechanisms. His research and development activities focus on the full custom hardware implementation of real-time sensory-motor systems using analog/digital neuromorphic circuits and emerging VLSI technologies.

Related Events (a corresponding poster, oral, or spotlight)