ESR6 - Spiking Neural Networks for Information Representation and Decoding

Using information-theoretic computational analyses of real spike trains, based on the concept of intersection information, we will determine how information about tactile stimuli is carried by neuronal populations, from peripheral to central at various levels, and how the representation at a given stage of processing is read out to give rise to progressively determining the final percept and its use for the task.

Overview

Objectives

Using information-theoretic computational analyses of real spike trains, based on the concept of intersection information, we will determine how information about tactile stimuli is carried by neuronal populations, from peripheral to central at various levels, and how the representation at a given stage of processing is read out to give rise to progressively determining the final percept and its use for the task. The metric for the biological validity of this approach (and the devices based upon it) is to use the candidate decoding algorithm in order to specify both the object being contacted by the sensory system and the subjects’ psychophysical choice. Correct decoding of the stimulus indicates that the decoding algorithms have identified information-carrying algorithms, while correct decoding of choice indicates that the algorithms have identified the same elements used by the brain to construct perception. The fellow will therefore develop psychophysical behavioral paradigms for rats and humans in parallel, with methods for fully characterizing motor strategy and sensory input. S/he will record neuronal population data sets at multiple stages of rat tactile processing pathway.

Expected Results

Description of the spike-timing based neural population codes employed for tactile coding and perceptual decisions across the brain. Mathematical extrapolation of these principles to rules to encode information in artificial sensors and to use this information in robots for performing tasks.

Secondments

UNIBI

developing SNN for hardware implementation
EPFL and SensArs

decoding mechanisms for sensory feedback in prosthetic devices

Supervisors

S. Panzeri
M. Diamond
E. Chicca
S. Micera
F. Petrini

Alejandro Pequeño Zurro

While there is still many unknowns in the research challenge of understanding general intelligence, latest trends point towards the idea that the brain is not an isolated entity and the input and processes from the environment are constantly modifying your brain. During my previous studies in the area of robotics I have been researching sensory-adaptive systems inspired by animal experiments. It was a natural step for me to transition towards computational neuroscience focus on sensory systems and give some light in how sensory processing in the brain modifies our behaviour.

Neutouch for me

I found this research project very interesting since it combines a wide approach between artificial systems, cognition and neuroscience which perfectly matches with my multidisciplinary background.

Info

Research Topics

Computational Neuroscience
Institution

Istituto Italiano di Tecnologia (Genova) — Scuola Internazionale Superiore di Studi Avanzati (SISSA)
Background

B.Sc. Telecommunication, Universidad Autonoma de Madrid (2010-2015)
M.Sc. Robotics, University of Southern Denmark (2016-2018)

Thesis

Exploiting spatio-temporal patterns with neuromorphic systems. SISSA, 2023.

Abstract

New demands in artificial intelligence, the increase of data available, and the forecasts of reaching Moore’s law ceiling push algorithms towards the edge for low latency, low power, and highly intelligent devices. Neuromorphic systems mimic brain-like computations to capture the efficiency and adaptive behaviour exhibited in biological systems being promising candidates to lead the new generation of artificial systems. Sensory information is encoded with precise spatio-temporal patterns in the nervous system. Similarly, neuromorphic computation utilises event-based representations in its computational systems, inspiring its use for building artificial cognitive systems. In this work, we build Spiking Neural Networks (SNNs) in the tactile and auditory sensory modalities for classification tasks in the context of full event-based sensory systems. Compared with standard machine learning implementations in GPU, SNNs implemented in neuromorphic hardware output close to 500 times less energy consumption highlighting the strengths of the new hardware paradigm. Also, we explore how to encode spatio-temporal features from sensing devices emphasising the benefits of using full neuromorphic event-based sensors and systems. We further explore improvements in spiking networks based on the Leaky-Integrate-and-Fire (LIF) models. The proposed network’s architecture implements Time Difference Encoders (TDEs), a cell model based on brain-inspired computations. We find promising results with a 92% reduction of synaptic operations and highly interpretable network results against typical current-based recurrent LIF networks. The results of this work contribute towards building neuromorphic sensing systems from sensor devices, algorithms and circuit design to quantify the strengths of the new computational paradigm with promising capabilities for intelligent machines on edge inspired by their biological counterparts.

Publications

Pequeño-Zurro, A., Khacef, L., Panzeri, S., & Chicca, E. (2025). Towards efficient keyword spotting using spike-based time difference encoders. arXiv preprint arXiv:2503.15402.

Panzeri, S., Janotte, E., Pequeño-Zurro, A., Bonato, J., & Bartolozzi, C. (2023). Constraints on the design of neuromorphic circuits set by the properties of neural population codes. Neuromorphic computing and engineering, 3(1), 012001.

Top