CogniGron Seminar: Eleni Vasilaki (Computer Science, University of Sheffield, UK) - "Sparse Reservoir Computing (SpaCe)"

Neuroscience has long been an inspiration for Artificial Intelligence and Machine Learning. In this talk I will present fundamental ideas about biological learning in frutflies, and how these are related to Machine Learning. Inspired by the architecture of small brains, and within the framework of Ecco State Networks, I will discuss the importance of neuron selectivity to specific stimuli. I will then introduce a threshold per reservoir neuron as an efficient mechanism to achieve sparseness in the neuronal representation. The threshold is adapted via a gradient rule on an error function structurally identical to threshold learning via backpropagation.  And yet, a simple mathematical analysis of its consequences for the specific architecture shows that it is differs from backpropagation and that it leads to neuronal selectivity. I will show in simulations that, within this context, our approach is advantageous in terms of performance versus imposing sparseness of weights via L1 norm. I will also discuss how such learning architectures can be exploited in the context of neuromorphic engineering.

Work with Luca Manneschi and Andrew Lin

More about Eleni
Eleni graduated with a Bachelor’s degree in Informatics and Telecommunications and a Master’s degree in Microelectronics from the University of Athens, before taking her DPhil (PhD) at Sussex, in Computer Science and Artificial Intelligence. From 2004 to 2006 she worked at the University of Bern and from 2007 to 2009 at the Swiss Federal Institute of Technology Lausanne (EPFL). In 2009 she joined the University of Sheffield as Lecturer, where she is Professor since 2016.