Grant: Discovery Grants Program - Individual
Université Laval
$41,000.00
Intertwined dynamics of neurons and synapses in large neural networks
FY 2024-25
Summary
This proposal outlines an interdisciplinary research program aiming to advance our understanding of brain dynamics through theoretical and computational neuroscience.
Collaboratively with neurobiologists, we will develop and refine theoretical and computational tools for analyzing, modeling, and predicting neuronal responses to various stimuli and internal activity. This initiative will leverage the expertise of students from physical, mathematical, and computational domains.
Our five-year strategy focuses on four primary objectives:
1. Develop an artificial recurrent neural network model incorporating neuromodulation and a biologically-inspired online learning rule for analyzing real neuronal activity data from pharmacological and learning protocols.
2. Create a dimensionality reduction method for large neural networks experiencing Hebbian-like synaptic plasticity, aiding in predicting their evolution and learning capabilities.
3. Formulate a growing neural network model to explain connection pattern formations and specific graph-theoretical feature emergence, emphasizing network effective rank reduction through maturation.
4. Refine the existing information-theoretical framework quantifying structure-function relationships in random neural networks to better reflect biologically plausible scenarios where synapses and neurons interdependently interact.
These goals are informed by findings from our prior NSERC grant, insights from the Sentinel North program, and the Neuro-CERVO Alliance. Our work aligns with the UNIQUE research center's objectives, fostering synergy between neuroscience and AI researchers. Among our notable developments are:
1. NeuroTorch, a Python library facilitating neuroscience-oriented machine learning through artificial recurrent neural networks and biologically plausible learning algorithms.
2. Computational pipelines transforming raw neuronal activity data from the CERVO Brain Research Center into a format suitable for time-series analysis or network modeling.
3. Dimensionality reduction techniques enabling predictions on neural networks' large-scale and long-term behavior, and a spectral graph theory-based framework unveiling a novel characteristic of biological neural networks - their effective low rank.
4. Stochastic and deterministic dynamical models accounting for neuronal kinetic parameter heterogeneity and random response fluctuations to stimuli, alongside a new information theory-based framework quantifying complex network structure-function relationships.
Further work is necessary to align our models with neurobiological requirements, particularly in integrating synaptic plasticity mechanisms like Hebbian rules, and understanding reward prediction error signals correlated with neuromodulator levels. Enhancements in these areas could significantly enrich our methods' interpretive power when applied to real neuronal data, and inspiring the creation of new biologically inspired AI systems.
Program
Discovery Grants Program - Individual
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
RGPIN-2024-06492
Name nom
Desrosiers, Patrick
Department département
Physique, génie physique et optique
Organization id
51
Institution établissement
Université Laval
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
41000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1505
Committee name en
Physics
Committee name fr
Physique
Area of application code
1200
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Advancement of knowledge
Area of application fr
Avancement des connaissances
Research subject code
3113
Research subject group en
Physics
Research subject group fr
Physique
Research subject en
Biophysics
Research subject fr
Biophysique
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Intertwined dynamics of neurons and synapses in large neural networks
Keywords
biological neural networks; artificial recurrent neural networks; synaptic plasticity; neuromodulation; biologically inspired learning rules; emergent phenomena; dynamical resilience; random matrices; stochastic processes; animal behavior
Application summary
This proposal outlines an interdisciplinary research program aiming to advance our understanding of brain dynamics through theoretical and computational neuroscience.
Collaboratively with neurobiologists, we will develop and refine theoretical and computational tools for analyzing, modeling, and predicting neuronal responses to various stimuli and internal activity. This initiative will leverage the expertise of students from physical, mathematical, and computational domains.
Our five-year strategy focuses on four primary objectives:
1. Develop an artificial recurrent neural network model incorporating neuromodulation and a biologically-inspired online learning rule for analyzing real neuronal activity data from pharmacological and learning protocols.
2. Create a dimensionality reduction method for large neural networks experiencing Hebbian-like synaptic plasticity, aiding in predicting their evolution and learning capabilities.
3. Formulate a growing neural network model to explain connection pattern formations and specific graph-theoretical feature emergence, emphasizing network effective rank reduction through maturation.
4. Refine the existing information-theoretical framework quantifying structure-function relationships in random neural networks to better reflect biologically plausible scenarios where synapses and neurons interdependently interact.
These goals are informed by findings from our prior NSERC grant, insights from the Sentinel North program, and the Neuro-CERVO Alliance. Our work aligns with the UNIQUE research center's objectives, fostering synergy between neuroscience and AI researchers. Among our notable developments are:
1. NeuroTorch, a Python library facilitating neuroscience-oriented machine learning through artificial recurrent neural networks and biologically plausible learning algorithms.
2. Computational pipelines transforming raw neuronal activity data from the CERVO Brain Research Center into a format suitable for time-series analysis or network modeling.
3. Dimensionality reduction techniques enabling predictions on neural networks' large-scale and long-term behavior, and a spectral graph theory-based framework unveiling a novel characteristic of biological neural networks - their effective low rank.
4. Stochastic and deterministic dynamical models accounting for neuronal kinetic parameter heterogeneity and random response fluctuations to stimuli, alongside a new information theory-based framework quantifying complex network structure-function relationships.
Further work is necessary to align our models with neurobiological requirements, particularly in integrating synaptic plasticity mechanisms like Hebbian rules, and understanding reward prediction error signals correlated with neuromodulator levels. Enhancements in these areas could significantly enrich our methods' interpretive power when applied to real neuronal data, and inspiring the creation of new biologically inspired AI systems.